ARTICLE

https://doi.org/10.1038/s41467-019-11568-0 OPEN competes with MiT/TFE in regulating lysosomal biogenesis and through an epigenetic rheostat

Ida Annunziata 1, Diantha van de Vlekkert 1, Elmar Wolf 2, David Finkelstein3, Geoffrey Neale 4, Eda Machado 1, Rosario Mosca 1, Yvan Campos1, Heather Tillman 5, Martine F. Roussel 6, Jason Andrew Weesner 1,7, Leigh Ellen Fremuth 1,7, Xiaohui Qiu 1, Min-Joon Han8, Gerard C. Grosveld 1 & Alessandra d’Azzo 1 1234567890():,;

Coordinated regulation of the lysosomal and autophagic systems ensures basal catabolism and normal cell physiology, and failure of either system causes disease. Here we describe an epigenetic rheostat orchestrated by c-MYC and histone deacetylases that inhibits lysosomal and autophagic biogenesis by concomitantly repressing the expression of the transcription factors MiT/TFE and FOXH1, and that of lysosomal and autophagy . Inhibition of histone deacetylases abates c-MYC binding to the promoters of lysosomal and autophagy genes, granting promoter occupancy to the MiT/TFE members, TFEB and TFE3, and/or the autophagy regulator FOXH1. In pluripotent stem cells and , suppression of lysosomal and autophagic function is directly downstream of c-MYC overexpression and may represent a hallmark of malignant transformation. We propose that, by determining the fate of these catabolic systems, this hierarchical switch regulates the adaptive response of cells to pathological and physiological cues that could be exploited therapeutically.

1 Department of Genetics, St. Jude Children’s Research Hospital, Memphis, TN 38105, USA. 2 Department of Biochemistry and Molecular Biology, Biocenter, University of Würzburg, Würzburg 97074, Germany. 3 Department of Computational Biology, St. Jude Children’s Research Hospital, Memphis, TN 38105, USA. 4 Hartwell Center, St. Jude Children’s Research Hospital, Memphis, TN 38105, USA. 5 Department of Pathology, St. Jude Children’s Research Hospital, Memphis, TN 38105, USA. 6 Department of Tumor Cell Biology, St. Jude Children’s Research Hospital, Memphis, TN 38105, USA. 7 Department of Anatomy and Neurobiology, College of Graduate Health Sciences, University of Tennessee Health Science Center, Memphis, TN 38163, USA. 8 Department of Hematology, St. Jude Children’s Research Hospital, Memphis, TN 38105, USA. Correspondence and requests for materials should be addressed to A.d. (email: [email protected])

NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11568-0

ukaryotic development and differentiation programs Another prototypical member of the b-HLH depend on the fine-tuned regulation of expression, class of transcription factors is c-MYC (hereafter referred to as E fi which is achieved by continuous remodeling/modi cation MYC), which also binds to E-boxes near the core promoter ele- of chromatin through reversible epigenetic marks on histone ments of target genes25,26. MYC functions as master regulator of tails1. The latter modulates the assembly/compaction of chro- cellular metabolism and proliferation27. Under physiological matin and the recruitment of transcription factors2. Among the conditions, MYC transcription is controlled by developmental modifications that change chromatin structure, histone acetyla- and mitogenic cues, and decreases in differentiated cells in tion or deacetylation of specific lysine residues by histone acet- absence of additional activating signals. The MYC is short ylases or deacetylases (HDACs) allows for the rapid adaptation of lived in proliferating cells, which adds an extra checkpoint to cells to extra- and intracellular signals3. In mammals, HDACs avoid transformation. In addition to promoting cell growth and comprise a family of 18 genes, which are grouped, based on their proliferation, MYC inhibits terminal differentiation of cells via homology to their yeast counterparts, into classes I to IV. Classes activation or repression of target genes. Both MYC functions I, II, and IV include 11 Zn2+-dependent family members, which depend on the recruitment of chromatin-modifying complexes, are referred to as “classical” HDACs4. In general, HDACs display including HDACs, which, by changing accessibility to the tran- limited substrate selectivity and rely on the association with scriptional machinery, allow or prevent MYC-driven transcrip- transcription factors or repressor complexes to attain their spe- tion of the corresponding genes28. Under physiological cificity at target DNA sites5. conditions this mechanism is strictly regulated, but in cancer cells Chromatin modifications by HDACs influence a multitude of constitutively elevated expression of MYC shifts the balance cellular pathways through repression of transcription of meta- toward abnormal activation or repression of MYC responsive bolic genes. Overexpressed or deregulated HDACs have been genes27,29. When overexpressed, MYC is a potent oncogene, and linked to human conditions mostly associated with aging, such as its constitutive high-expression drives many tumor types and is cancer, diabetes, cardiac hypertrophy, and neurodegenerative often associated with cancer aggressiveness and poor prognosis29. diseases6–8. Inhibition of HDACs promotes growth arrest and cell Here we define a previously unknown antagonistic role differentiation or apoptosis9. For this reason, the Food and Drug between HDAC/MYC and MiT/TFE in the epigenetic and tran- Administration (FDA) has approved HDAC inhibitors for the scriptional control of the two-major cellular catabolic machi- treatment of various malignancies, in which they cause altered neries. We further unveil that inhibition of HDACs and reduction expression of only a small subset of genes, suggesting that acet- of MYC levels displace MYC from E-Box promoter sites of ylation/deacetylation is restricted to specific chromatin regions10. lysosomal and autophagy genes, which grants occupancy of the HDAC inhibitors have also been used to reverse disease- same sites to MIT/TFE members and results in activation of associated epigenetic states in adult cardiovascular, neurodegen- lysosomal biogenesis and autophagy. Finally, we show that this erative and inflammatory diseases, and more recently in some of competing transcriptional rheostat is established during cell the pediatric lysosomal storage diseases (LSDs)11–14. reprogramming and differentiation in both normal and disease control the breakdown, processing or recycling of conditions. long-lived , nucleic acids, carbohydrates, and complex lipids that reach the organelles through the biosynthetic, endo- cytic, phagocytic or autophagic route15. These processes are Results carried out by a large range of hydrolytic , membrane HDACs regulate lysosomal biogenesis and MiT/TFE members. proteins, transporters, and ion channels that are ubiquitously but To investigate whether epigenetic marks regulate lysosomal differentially expressed in different cell types and tissues. The function, we tested whether suppression of histone acetylation importance of a fully functional lysosomal system in maintaining with suberoylanilide hydroxamic acid (SAHA)30, an FDA cell is evidenced in the complex pathobiology of approved HDAC pan-inhibitor, altered the expression of lyso- LSDs. Endo-lysosomal accumulation of undigested or partially somal genes. Microarray expression profiles, obtained from either digested substrates in cells of virtually all organs is the hallmark of DMSO- or SAHA-treated HeLa cells, identified differentially these diseases and determines phenotypic penetrance and out- activated or suppressed pathways by gene set enrichment analysis come. The lysosomal system is strictly connected to the autop- (GSEA) and database for annotation, visualization, and integrated hagic system because autophagosomes need to fuse with discovery (DAVID) analysis (Fig. 1a, Supplementary Fig. 1a, and lysosomes to execute the degradative stage of the pathway16. Supplementary Table 1 and Supplementary Data 1). Therefore, impaired lysosomal activity leads also to the accu- Activation of the lysosomal pathway was among the top- mulation of autophagic substrates and autophagic disfunction17. ranked responses by GSEA and DAVID in SAHA-treated cells Autophagy is a highly conserved catabolic process mainly acti- (Fig. 1a and Supplementary Table 1 and Supplementary Data 1), vated during starvation, allowing cells to generate energy during while the most suppressed pathways were related, as previously nutrient deprivation. reported31, to cell cycle and mitotic division (Supplementary Both the lysosomal and autophagic systems have recently Data 1). HDAC inhibition in HeLa cells increased the mRNA gained attention as modulators of nutrient sensing and abundance of 19 out of 22 lysosomal genes tested (Fig. 1b); signaling17,18. The constituents of this organellar network are among them the sialidase-encoding gene 1 transcriptionally regulated by members of the microphthalmia- (NEU1) was the top responder (Fig. 1b). To assess the generality associated family of basic helix loop helix (b-HLH) leucine zipper of these findings, we also tested the effects of SAHA treatment in transcription factors (MiT/TFE), which include TFEB, TFEC, additional human cancer cell lines (RH30, rhabdomyosarcoma TFE3, and MITF19,20. When altered, MiT/TFE members induce and Sy5y, neuroblastoma), and in primary skin fibroblasts. The cancer and other severe diseases21. All MiT/TFE transcription results were comparable to those obtained with HeLa cells factors recognize a unique enhancer box (E-box) DNA motif (Supplementary Fig. 1b–d). In addition to the data presented [also referred to as CLEAR (coordinated lysosomal expression here, we also confirmed activation of global lysosomal gene and regulation)] within the proximal promoters of lysosomal expression by querying publicly available datasets obtained from and autophagy genes20,22,23, and regulate cellular catabolism and endothelial-, transformed lymphoblastoid- and breast cancer-cells nutrient-dependent lysosomal response19,24. treated with SAHA (Supplementary Fig. 1e–g). The strong

2 NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11568-0 ARTICLE

a 0.45 b 64.0 HeLa: RT-qPCR 0.40 ** 0.35 32.0 0.30 ** 0.25 16.0 ** 0.2 ** 8.0 ** 0.15 ** 0.10 0.05 4.0

Enrichment score (ES) 0.00 –0.05 2.0 1.0 relative to DMSO 0.5

Normalized mRNA expression 0.25 0 5000 10,000 15,000 20,000 25,000 Top ranking genes IDS GLA GBA GNS TPP1 FUCA1 LIPA MCOLN1 GBA CTSK GLB1 NEU1 NPC1 CTSA ACP2 CTSB HEXA ARSA PSAP CTSD ARSB SGSH LAMP1 LAMP2 CTSL2 ATP6V0A1 MAN2B1 IDS AP1B1 MFSD8 MCOLN1

ARSA AP3B2 NPC2 AP1S2 LGMN ATP6V0E1 NEU1 LAMP3 PLA2G15 GM2A CLN5

c def p = 0.0177 g h p = 0.0012 p < 0.0001 p < 0.001 12,000 25 30 p < 0.0001 2000 150 500 p < 0.0001 20 400 1500 8000 20 100 15 300 1000 10 200 4000 10 50 500 Lysosomal count 5 100 -HEX activity (nmol/mg/h) -GAL activity (nmol/mg/h) -MAN activity (nmol/mg/h) Lysosomal volume (MFI) NEU1 activity (nmol/mg/h) β β 0 0 0 0 α 0 0 DMSO SAHA DMSO SAHA DMSO SAHA DMSO SAHA DMSO SAHA DMSO SAHA HeLa HeLa HeLa HeLa HeLa HeLa

i j k p = 0.0014 l m p = 0.0053 n p < 0.0001 p = 0.0011 p = 0.0008 100 25 2500 200 p = 0.001 600 100

80 20 2000 80 150 60 15 1500 400 60 100 40 10 1000 40 50 200 20 5 500 20 -GAL activity (nmol/mg/h) -HEX activity (nmol/mg/h) -MAN activity (nmol/mg/h) CA activity (nmol/mg/min) CA activity (nmol/mg/min) NEU1 activity (nmol/mg/h) β β 0 0 0 α 0 0 0 DMSO SAHA DMSO Romi DMSO Romi DMSO Romi DMSO Romi DMSO Romi HeLa HeLa HeLa HeLa HeLa HeLa o p 2.5 HeLa: ChIP HDAC2 DMSO ABCA2 ACP2 ARSB ASAH1 ARSA ARSG TMEM92 ABCB9 AGA ACP5 CLCN7 CLN3 CTNS CTSF CTSH ENTPD4 FUCA1 GALNS HEXB HYAL1 IDUA LMBRD1 NAGA NAGLU OSTM1 PCYOX1 PPT1 RNASET2 SGSH SMPD1 TFE3 CCZ1 CTBS CTSD GM2A HPSE LGMN NPC2 CPVL CTSC DNASE2 GALC GBA HEXA HGSNAT LAMP1 MAN2B2 NAAA NEU1 NEU4 PLBD2 PPT2 TFEB TPP1 CTSK GLA IDS TFEC CD63 CD68 CLN5 CREG1 CTSA CTSB CTSL CTSZ EPDR1 GAA GGH GLB1 GNS GUSB HYAL2 IFI30 LAPTM44 LAPTM5 LIPA LITAF MAN2B1 MANBA MCOLN1 MFSD8 MITF NAGA NCSTN NEU3 NPC1 PLA2G15 PSAP SCARB2 SCPEP1 SIAE SLC17A5 SLC35F6 SLC36A1 TMEN9 CTSS LAMP2 LAMP3 MPO NEU2 TMEM55B TMEM74 DMSO IgG +5 K Value

Color key SAHA 0 0.4 0.8 2.0 SAHA IgG TSS

–5 K 1.5 +5 K 1.0 % of input

–5 K 0.5 +5 K 0 NEU1 LAMP1 MCOLN1 TFEB TFE3 NEU1 INT

–5 K +5 K q HeLa ChIP: Acetyl-H3K14 * * TSS 6.0 * DMSO –5 K DMSO IgG +5 K SAHA SAHA IgG

TSS4.0 TSS TSS –5 K * +5 K % of input

H1ESC HEPG2 K562 2.0 TSS * –5 K

0 NEU1 LAMP1 MCOLN1 TFEB TFE3 NEU1 INT induction of lysosomal biogenesis was further proven by the details are beyond the scope of the present study, the difference expansion of the lysosomal compartment (Fig. 1c, d), detected by between NEU1 mRNA/protein levels and NEU1 enzymatic flow cytometry of lysotracker green positive lysosomes, the activity observed upon SAHA treatment could be explained by increased levels of various membrane and soluble lysosomal the low-specific activity of NEU1 towards the synthetic substrate components (Supplementary Fig. 2a–f), and the enhanced and/or by the rate limiting amount of PPCA (Protective Protein/ enzymatic activity of glycosidases, including NEU1, and cathe- Cathepsin A) available for chaperoning and activating NEU1 in psins (Fig. 1e–i and Supplementary Fig. 2g–i). While further lysosomes32.

NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11568-0

Fig. 1 HDACs epigenetically regulate the expression of lysosomal genes. a GSEA demonstrated significant activation of the lysosomal pathway in HeLa cells treated with SAHA (20 μM for 24 h). The top 20 upregulated genes are shown at the bottom of the plot. b Expression analysis of lysosomal genes after SAHA treatment (20 μM for 24 h). The box and whisker plot show normalized expression of the lysosomal mRNAs in SAHA-treated HeLa cells relative to that in DMSO-treated cells. n = 5 biologically independent samples. c Lysosomal volume was measured by FACS analysis as mean fluorescence intensity (MFI) after lysotracker staining of HeLa cells treated with SAHA (20 μM for 24 h) and compared to cells treated with DMSO, n = 9 biologically independent samples. d Lysosomal count of transmission electron microscopy images n = 20 images. e–n Activity assays for NEU1 (n = 7 biologically independent samples), β- (β-HEX) (n = 4 biologically independent samples), α- (α-MAN) (n = 5 biologically independent samples), β-galactosidase (β-GAL) (n = 5 biologically independent samples) and cathepsin A (CA) (n = 5 biologically independent samples) in HeLa cells treated with DMSO or SAHA (e–i) (20 μM for 24 h), or in HeLa cells treated with DMSO or romidepsin (j-n) (romidepsin, Romi, 10 nM for 24 h; n = 6 biologically independent samples). o HDAC2 occupancy was analyzed using ChIP-seq datasets available at the ENCODE/Haib project. Genes shown on the left of the vertical black line are occupied by HDAC2. p = 4.65e-11, odds ratio = 4.4, Fisher’s exact test. p ChIP analysis of the promoters of NEU1, LAMP1, MCOLN1, TFEB, and TFE3 using anti-HDAC2 antibody in HeLa cells treated with SAHA (20 μM for 24 h) or DMSO (n = 3 biologically independent samples). Oligos encompassing a genomic region at +5 Kb from the TSS of NEU1 (NEU1 INT) were used as control for non-specic antibody binding (n = 3 biologically independent samples); IgG control n = 3 biologically independent samples. q ChIP analysis of the promoters of NEU1, LAMP1, MCOLN1, TFEB, and TFE3 and NEU1 INT using acetyl histone H3 Lys 14 antibody (Acetyl-H3K14) in HeLa cells treated with SAHA (20 μM for 24 h) or DMSO (n = 8 biologically independent samples); IgG control n = 3 biologically independent samples. Boxes represent the mean value and bar inside the box represents median value; upper bar represents maximum of distribution; lower bar represents minimum of distribution (95% confidence level). Graphs shown in (c–n) and (p, q) are presented as mean ±SD. Statistical analysis was performed using Student t-test

In order to narrow the pan effect of SAHA inhibition on intriguing because it has been well documented that MYC tran- HDACs, we also tested in the same experimental setup the effects scription and protein levels are directly modulated by HDAC of romidepsin, a more potent and selective class I HDAC activity28,36,37 and that MYC and HDACs interact38,39.Inlinewith inhibitor33. The activation of lysosomal (Supple- these observations we showed that silencing of HDAC2 drastically mentary Fig. 3a) and consequent increase in the activity of several reduced MYC protein levels (Fig. 2b, c and Supplementary Fig. 4a, lysosomal enzymes (Fig. 1j–n) closely recapitulated the results b), that MYC and HDAC2 co-immunoprecipitated (Fig. 2d, e and with SAHA, strongly supporting the idea that members of class I Supplementary Fig. 2c, d) and that HDAC2 was bound to the MYC HDACs are primarily responsible for the epigenetic suppression promoter (Fig. 2f). We noticed that the E-box motif recognized by of lysosomal biogenesis. MYC25 remarkably overlaps with the CLEAR motif recognized by We next investigated whether HDACs engage the promoters of TFEB and TFE3, raising the possibility that MYC binds the pro- lysosomal genes by querying publicly available ChIP datasets for moters of lysosomal genes. To test this hypothesis, we queried the class I HDACs. For this analysis we used HDAC2 as a ChIP-seq datasets performed with anti-MYC antibody29,40 and proxy (http://genome.ucsc.edu/cgi-bin/hgTrackUi?db=hg19&g= found that MYC occupied not only the promoters of lysosomal wgEncodeHaibTfbs), as it is one of the most widely expressed genes (Fig. 2g, h and Supplementary Table 2 and Supplementary nuclear HDACs34. We found that most (82 of 102) lysosomal Data 2) but also those of MiT/TFE family members TFEB and genes were overrepresented among the HDAC2-bound targets TFE3 (Fig. 2i and Supplementary Fig. 4e, Supplementary Data 2 (Fig. 1o). These results were confirmed by ChIP performed with and Supplementary Table 3). In addition, ChIP analyses of HeLa HeLa cells treated with either SAHA or DMSO using an anti- cells, treated or not with SAHA, confirmed that in untreated cells HDAC2 antibody (Fig. 1p). This assay also revealed that HDAC2 MYC occupied the promoters of TFEB and TFE3, but was dis- occupied the promoters of both TFEB and TFE3 (Fig. 1p), two of placed upon inhibition of HDACs (Fig. 2j). We also observed that the MiT/TFE members known to regulate lysosomal function and MYC was bound to its own promoter in control cells, but this metabolism20,22. It is important to notice that inhibition of binding was reduced in SAHA-treated cells (Fig. 2k). Furthermore, HDAC2 with SAHA did not alter its binding capacity to the chromatin precipitated with the MYC antibody and sequentially promoters; this is because SAHA specifically affects the histone precipitated with HDAC2 antibody revealed that TFEB and TFE3 deacetylase activity of HDACs without altering their protein promoters were co-occupied by MYC and HDAC2 (Fig. 2l). levels35. Remarkably, silencing of only HDAC2 (Supplementary In agreement with these observations, MYC mRNA and Fig. 3b, c) was sufficient to increase the activity of lysosomal protein levels were significantly downregulated upon treatment enzymes (Supplementary Fig. 3d–g) in a manner comparable to with HDAC inhibitors (Fig. 3a, b and Supplementary Fig. 4a, b that obtained upon HDAC inhibition. Activation of gene and Supplementary Fig. 4f–h). In contrast, the expression of the transcription by inhibiting HDACs was also measured by MiT/TFE members was significantly increased upon SAHA/ increased acetylation of histone 3 (H3) on lysine 14 (H3K14) of romidepsin treatment, albeit in a cell-specific manner, which is the promoter regions of several lysosomal genes as well as of likely due to the relative abundance of these transcription factors TFEB and TFE3 genes (Fig. 1q). Together these results indicate in different cell types: TFE3 was increased in HeLa cells (Fig. 3a, that HDACs, and specifically HDAC2, epigenetically control the b), MITF, TFEB, and TFE3 were all increased in RH30 expression levels not only of a plethora of lysosomal genes but (Supplementary Fig. 4f) and Sy5y (Supplementary Fig. 4g) cell also of the MiT/TFE transcription factors. lines, while MITF was increased exclusively in skin primary fibroblasts (Supplementary Fig. 4h). Performing ChIP assays of MYC represses lysosomal biogenesis. In search for putative HeLa cells treated with SAHA, we further demonstrated that binding sites in the promoters of lysosomal MYC downregulation enabled binding of TFEB and TFE3 to the genes bound by HDAC2, we performed motif analysis and iden- promoters of lysosomal genes and also to their own respective tified the E-box as the motif with the highest probability of occu- promoters (Fig. 3c–e). This mechanism may allow for robust but pancy. E-box binding sites are recognized by the b-HLH family of tightly controlled activation of lysosomal gene expression, as transcription factors (Fig. 2a) that include MiT/TFE members and consequence of acetylation of histones. MYC, the master regulator of metabolism27, The potential To unequivocally demonstrate that TFEB and TFE3 are the engagement of MYC at lysosomal gene promoters was particularly primary drivers of lysosomal biogenesis under SAHA treatment,

4 NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11568-0 ARTICLE

a b shHDAC2 shHDAC2 c Motif/Name Rank P-value kDa E1 2 E1 2 250 1.5 1 1e–11 150 E-box(HLH)/Promotor/Homer 100 1.0 *

2 1e–7 75 0.5 **** Usf2(HLH)/C2C12-Usf2-ChiPSeq(GSE36030)/Homer 50 protein levels Normalized MYC 0 3 1e–5 37 USF1(HLH/GM12878-Usf1-ChiPSeq(GSE32465)/Homer Empty 25

Anti-MYC Coomassie shHDAC2.1 shHDAC2.2

d e f HeLa: ChIP HDAC2 IP anti-MYC IP-IgG Beads Beads IP-IgG IP anti-MYC 3.0 p = 0.0109

DMSO DMSO IgG SAHA DMSO SAHA SAHA SAHA SAHA DMSO DMSO DMSO DMSO DMSO SAHA 2.0 SAHA 75 75 SAHA IgG

50 * * * * % Input 50 1.0 IgG IgG 37 37 WB:anti-HDAC2 WB: anti-MYC 0 MYC g h i 2 kb 5 kb

y n y Inp Myc Inp Myc 10 kb y Inp Myc 70 Inp Myc 50 80 Tags Tags 70 Tags 50 80

Neu1 Lamp1 Tfeb TSS TSS TSS

2 kb 2 kb Inp Myc

2 kb Inp Myc 120 80 40 Tags Tags 80 Tags 120 40

Ctsa Mcoln1 Tfe3 TSS TSS TSS

j k HeLa: ChIP MYC l HeLa: ChIP MYC re-ChIP HDAC2 2.5 HeLa: ChIP MYC DMSO 1.5 2.0 0.2 * DMSO IgG p = 0.0104 DMSO DMSO SAHA IgG IgG SAHA IgG 1.25 2.0 DMSO DMSO IgG p = 0.0116 p = 0.03 1.0 SAHA 1.5 **** SAHA IgG ** 0.75 1.0 0.1 1.0 ** % of input % of input % of input 0.5 % of input

0.5 0.25

0 0 0 0 NEU1 LAMP1 MCOLN1 TFEB TFE3 NEU1 INT MYC TFEB TFE3 we tested the effects of this inhibitor in murine embryonic we observed a marked reduction in lysosomal gene expression fibroblasts (MEFs) that had CRISPR-mediated ablation of both (Fig. 3f, g) and reduced NEU1 activity (Fig. 3h) compared to WT transcription factors (dKO)41 (Supplementary Fig. 4i, j). In spite MEFs. Together these results confirm that MIT/TFE members are of the complete absence of these transcription factors, we still indeed responsible for the activation of lysosomal genes after measured activation of lysosomal biogenesis upon SAHA HDAC inhibition. treatment of these dKO cells (Fig. 3f, g). We found that these To further understand how MYC fine-tunes lysosomal unexpected results were likely due to the upregulation in the dKO biogenesis, we tested a set of mouse group 3 medulloblastoma cells of Mitf expression (Supplementary Fig. 4k). This finding is in cell lines overexpressing Myc40,44. In those cells, the expression of agreement with the notion that members of the Mit/Tfe family murine Mit/Tfe members was significantly downregulated, as was function cooperatively or redundantly in the regulation of that of all tested lysosomal genes (Fig. 4a, b and Supplementary lysosomal biogenesis42,43. Upon shRNA silencing of Mitf in the Data 2). The latter was accompanied by a reduction of the dKO cells, followed by SAHA treatment (Supplementary Fig. 4k), lysosomal volume (Supplementary Fig. 4l), decreased lysosomal

NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11568-0

Fig. 2 MYC occupies the promoters of lysosomal genes and that of TFEB and TFE3. a Motif analysis using HDAC2-binding sites present in lysosomal genes. b Left, silencing of HDAC2 downregulated MYC protein expression in HeLa cells. Right, Coomassie stained immunoblot used as the loading control. c Quantification of MYC levels in HDAC2 silenced HeLa cells normalized to loading control (n = 3 independent experiments). d Co-immunoprecipitation of MYC with HDAC2 from lysates obtained from HeLa cells. HDAC2 band is labeled with an asterisk. Heavy chain IgG is labeled with an arrow. e Immunoprecipitation of MYC followed by immunoblot of MYC protein. MYC band is labeled with an asterisk. Heavy chain IgG is labeled with an arrow. f Graph represents HDAC2 binding to the promoter of MYC (n = 5 biologically independent samples) in HeLa cells treated with DMSO and SAHA (20 μM for 24 h); IgG control (n = 3 biologically independent samples). g–i Myc binding to the promoters of (g) Neu1, Ctsa,(h) Lamp1, Mcoln1 and i Tfeb and Tfe3 was analyzed in ChIP-seq datasets performed with anti-Myc antibody in mouse group 3 medulloblastoma cells overexpressing Myc (Trp53–/– overexpressing Myc). Input DNA (Inp) serves as reference. j, k Histograms represent MYC binding to the promoters of (j) NEU1 (n = 3 biologically independent samples), LAMP1 (n = 3 biologically independent samples), MCOLN1 (n = 3 biologically independent samples), TFEB (n = 6 biologically independent samples), TFE3 (n = 3 biologically independent samples) and (k) MYC (n = 3 biologically independent samples) in HeLa cells treated with SAHA (20 μM for 24 h) or DMSO. NEU1 INT oligos were used as negative control for non-specific antibody binding (n = 3 biologically independent samples); IgG control (n = 3 biologically independent samples). l Sequential ChIP experiments were performed from HeLa cells with anti-MYC antibody followed by anti-HDAC2 antibody and analyzed by RT-qPCR at the promoter region of TFEB and TFE3 (n = 3 biologically independent samples); IgG control (n = 3 biologically independent samples). All the graphs are presented as mean ± SD. Statistical analysis was performed using the Student t-test. *p < 0.05, **p < 0.01, ****p < 0.0001

Neu1 activity and decreased Lamp1 protein levels results were again confirmed by ChIP assays with anti-MYC (Supplementary Fig. 4m–o). In contrast, computational analysis antibody showing the binding of MYC to the promoter of the of U2OS osteosarcoma cells silenced for MYC29, showed a MAP1LC3B gene in SAHA-treated HeLa cells (Fig. 6b). These transcriptional activation of the MIT/TFE members and of observations suggest a model in which MYC activation represses lysosomal genes (Fig. 4c, d). To corroborate these findings, we autophagy genes. In fact, this is the case because medulloblastoma tested lysosomal gene expression in HeLa cells silenced for MYC cells overexpressing Myc40 showed downregulated expression of (Supplementary Fig 4p, q). We demonstrated that merely autophagy genes (Fig. 6c), resulting in a decreased LC3II/LC3I silencing MYC already induced the expression of lysosomal ratio (Fig. 6d, e and Supplementary Fig. 6a). In contrast, querying genes that was further enhanced upon HDAC inhibition (Fig. 4e). the U2OS osteosarcoma dataset29 showed enhanced transcription This resulted in increased enzymatic activity of NEU1 (Fig. 4f). of several autophagy genes (Fig. 6f). Consistent with these results, silencing of MYC in HeLa cells led to a statistically significant The HDAC-MYC repressor rheostat regulates autophagy. increase in the expression of several autophagy genes, which was Considering the high degree of crosstalk between the lysosomal further enhanced by SAHA treatment (Fig. 6g). and the autophagic systems, we explored whether autophagy is Altogether, these data underscore that HDACs and MYC also controlled by the HDAC/MYC axis. Inhibition of HDACs cooperate in suppressing not only lysosomal biogenesis but also significantly increased the expression of many autophagy genes in autophagy. both primary fibroblasts and cancer cell lines (Fig. 5a, Supple- mentary Fig. 5a–e). These genes encode proteins that orchestrate the induction, formation and maturation of autophagosomes and FOXH1 regulates autophagy. To uncover the transcriptional the selection, expansion and degradation of their cargo45. Upon machinery cooperating with HDACs in the regulation of autop- fi SAHA treatment, analysis of the lipidated LC3 protein and the hagy, we performed motif-discovery analysis and identi ed the LC3II/LC3I ratios (Supplementary Fig. 5f–k), which specify TGT[GT][GT]ATT motif, which is bound by the FOXH1 tran- fi autophagosomes46, confirmed the induction of autophagy. In scription factor (Fig. 7a). FOXH1 was rst described as a devel- fi addition, the appearance of GFP-LC3B puncta (Fig. 5b, c) and the opmental gene, because it induces mesoderm speci cation in β increase in autophagic flux, assessed in cells treated or not with cooperation with activin, a member of the TGF- family. In β 47 the inhibitor bafilomycin A1, (Fig. 5d, e and Supplementary mammals, FOXH1 also mediates TGF- -type signaling . Fig. 5l), proved the occurrence in SAHA-treated cells of autop- To investigate if autophagy was induced by FOXH1, we hagosome synthesis and delivery of autophagic substrates to the overexpressed it in HeLa cells (Supplementary Fig. 6b) and for degradation. showed that the mRNA levels of many autophagy genes were fi We next queried the ENCODE/Haib dataset for engagement of signi cantly increased (Fig. 7b). This was paralleled by activation HDAC2 with the promoters of autophagy genes. Most (45 of 53) of autophagy, as measured by the lipidation of LC3B (Fig. 7c, d of the autophagy genes analyzed were indeed occupied by and Supplementary Fig. 6c). We next questioned whether HDACs HDAC2 (Fig. 5f). ChIP assays of chromatin isolated from HeLa modulate the levels of FOXH1. We demonstrated that SAHA cells with anti-HDAC2 antibody showed that HDAC2 was bound treatment of HeLa, RH30 and Sy5y cells, but not of primary fi to the promoter of the autophagy gene MAP1LC3B (Fig. 5g). In human broblasts, induced FOXH1 mRNA expression (Fig. 7e – fi contrast, inhibition of HDAC activity by treatment with SAHA and Supplementary Fig. 6d f), suggesting a cell type-speci c led to increased acetylation of H3K14 histone mark on the activation of FOXH1 in response to SAHA. Enhanced FOXH1 MAP1LC3B promoter region (Fig. 5h). Notably, transcription of transcription was accompanied by the increase of FOXH1 protein autophagy genes was reduced upon SAHA treatment in MEFs in HeLa cells treated with SAHA (Fig. 7f, g). In addition, by dKO for Tfeb and Tfe3 and silenced for Mitf, suggesting that querying MYC-ChIP datasets derived from U2OS cells with 29 fi these transcription factors cooperate in the induction of induced MYC expression , we con rmed MYC binding to the autophagy upon SAHA treatment (Fig. 5i, j). promoter of FOXH1, which resulted in reduced expression of To further test the link between autophagy and MYC FOXH1 (Fig. 7h, i). Similarly, overexpression of Myc in mouse activation, we also interrogated Myc-ChIP-seq binding medulloblastoma cells negatively regulated Foxh1 expression 29 datasets29,40 from group 3 mouse medulloblastoma and found (Fig. 7j). In contrast, silencing of MYC in U2OS cells resulted in that the promoters of several autophagy genes were indeed the enhanced transcription of FOXH1 (Fig. 7k). Based on these occupied by Myc (Fig. 6a and Supplementary Data 3). These results, we propose that FOXH1 is an additional activator of

6 NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11568-0 ARTICLE

abHeLa RT-qPCR HeLa RT-qPCR 2.5 3.0 DMSO DMSO **** SAHA *** Romidepsin 2.0 2.0 1.5 *

1.0 1.0 ** 0.5 *** Normalized mRNA expression Normalized mRNA expression 0 0 MYC MITF TFEB TFE3 MYC MITF TFEB TFE3

c 1.5 HeLa: ChIP: TFEB * DMSO 1.25 DMSO IgG SAHA SAHA IgG 1.0

0.75

% of input 0.5 ** * 0.25

0 NEU1 LAMP1 MCOLN1 TFEB TFE3 NEU1 INT

d 2.0 HeLa: ChIP TFE3 e 12.0 HeLa: ChIP TFE3 DMSO DMSO ** DMSO IgG 10.0 p = 0.004 DMSO IgG 1.5 SAHA SAHA SAHA IgG SAHA IgG 8.0

1.0 * 6.0 % of input % of input 4.0 0.5 2.0

0 0 LAMP1 MCOLN1 TFEB TFE3 NEU1 NEU1 INT

f g MEFs: RT-qPCR h 12.0 * 4.0 100 * * * 80 9.0 3.0

60 6.0 2.0 40 relative to DMSO 3.0 relative to DMSO 1.0 20 NEU1 activity (nmol/mg/h) Normalized mRNA expression Normalized mRNA expression 0 0 0 DS DS DS WT + shC WT + shC WT + shC WT + shC dKO + shC dKO + shC dKO + shC dKO + shC dKO + shMitf dKO + shMitf dKO + shMitf dKO + shMitf WT + shC Neu1Arsa Lamp1 Mcoln1 dKO + shC dKO + shMitf autophagy, specifically regulated by HDAC and MYC that may sialidosis, a rare neurosomatic LSD caused by mutations in the control this process in a temporal and cell/tissue-specific manner. NEU1 gene49. Sialidosis is an orphan disorder for which there is currently no cure49. Sialidosis patients are usually classified based on the age of onset, severity of the symptoms and levels Functional effects of HDAC inhibition in the LSD sialidosis. of NEU1 residual activity in type I (less severe) and type II Given that HDAC inhibitors were shown to reverse some of the (more severe)49. Patient-derived fibroblasts treated with SAHA phenotypes in the LSDs Niemann Pick type C and Gaucher showed increased levels of their mutant NEU1 mRNA, albeit to disease11,13,48, we tested the potential therapeutic effects of a different extent (Fig. 8a). Immunoblots probed with anti- HDAC inhibition in fibroblasts derived from patients with NEU1 antibody revealed a marked increase of the mutant

NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11568-0

Fig. 3 MYC antagonizes MiT/TFE members. a, b RT-qPCR of MYC, MITF, TFEB, and TFE3 in (a) SAHA-treated (20μM for 24 h; n = 5 biologically independent samples) or b romidepsin-treated (10 nM for 24 h) HeLa cells (n = 3 biologically independent samples) and compared to DMSO-treated cells. c ChIP analyses of NEU1 (n = 3 biologically independent samples), LAMP1 (n = 3 biologically independent samples), MCOLN1 (n = 6 biologically independent samples), TFEB (n = 3 biologically independent samples), and TFE3 (n = 9 biologically independent samples) promoters using anti-TFEB antibody in HeLa cells treated with SAHA (20 μM for 24 h) or DMSO. NEU1 INT oligos were used as negative control for non-specific antibody binding (n = 3 biologically independent samples); IgG control (n = 3 biologically independent samples). d, e ChIP analyses of the promoters of (d) LAMP1 (n = 3 biologically independent samples), MCOLN1 (n = 3 biologically independent samples), TFEB (n = 7 biologically independent samples), TFE3 (n = 6 biologically independent samples) and (e) NEU1 (n = 3 biologically independent samples) using an anti-TFE3 antibody in HeLa cells treated with SAHA (20 μM for 24 h) or DMSO. NEU1 INT oligos were used as negative control for non-specific antibody binding (n = 3 biologically independent samples); IgG control (n = 3 biologically independent samples). f, g RT-qPCR of (f) Neu1 (g) Arsa, Lamp1, and Mcoln1 was performed in mouse embryonic fibroblasts (MEFs), in which Tfeb and Tfe3 were knocked out via CRISPR-technology (dKO) and in dKO with Mitf silencing (dKO + shMitf) (n ≥ 6 biologically independent samples) and treated with SAHA (20μM for 24 h) or DMSO. shC refers to cells transduced with shRNA control lentivirus. Statistical analysis was performed comparing the normalized expression of the lysosomal genes in MEFs WT + shC versus dKO + shC and MEFs WT + shC versus dKO + shMitf. h Activity assay for NEU1, in MEFs WT, Tfeb and Tfe3 dKO transduced with a sh lentivirus control (ShC) or with a lentiviral vector targeting Mitf (dKO + shMitf) (n = 3 biologically independent samples). MEFs of different genotypes were treated with DMSO and SAHA (20 μM for 24 h). Statistical analysis was performed comparing the activity of NEU1 after SAHA treatment in MEFs WT + shC to dKO + shC and dKO + shMitf. Boxes represent the mean value and bar inside the box represents median value; upper bar represents maximum of distribution; lower bar represents minimum of distribution (95% confidence level). Graphs are presented as mean ± SD. Statistical analysis was performed using the Student t-test. *p < 0.05, **p < 0.01, ****p < 0.0001

a bc MB: RT-qPCR U2OS (siMYC):RT-qPCR 64.0 MB: RT-qPCR ***** **** 2.5 **** 1.0 16.0 2.0 **** 4.0 * **** 1.5 0.1 ** 1.0 **** 1.0 0.25 relative to control relative to control **** relative to siControl 0.5 0.06 0.01 Normalized mRNA expression Normalized mRNA expression

Normalized mRNA expression 0 IDS GLA GBA GNS MYC MITF TFE3 GLB1 TFEB CTSA NEU1 NPC1 ACP2 CTSB HEXA ARSA PSAP CTSD ARSB SGSH

MYC LAMP1 MITF LAMP2 TFE3 MFSD8 TFEB MCOLN1 ATP6V0E1

d U2OS (siMYC):RT-qPCR e 64.0 HeLa shMYC: RT-qPCR f 4.0 125 32.0 **** **** **** p = 0.0135 3.0 **** 16.0 **** 100 **** **** **** 8.0 75 **** **** 2.0 4.0 * *** **** 50 * ** ** * * 2.0 25 relative to DMSO empty

relative to siControl 1.0 1.0 NEU1 activity (nmol/mg/h) Normalized mRNA expression 0.5 0 Normalized mRNA expression

0 GBA NEU1 CTSA CTSB HEXA ARSA Empty SAHA LAMP1 Empty DMSO Empty SAHA Empty SAHA Empty SAHA Empty SAHA shMYC SAHA Empty DMSO Empty DMSO Empty DMSO Empty DMSO shMYC DMSO shMYC SAHA shMYC SAHA shMYC SAHA shMYC SAHA shMYC DMSO shMYC DMSO shMYC DMSO shMYC DMSO ARSA LAMP1 MCOLN1 NEU1

Fig. 4 MYC represses lysosomal gene expression. a, b RT-qPCR of (a) Myc, Mitf, Tfeb, Tfe3 and (b) lysosomal genes was performed in mouse group 3 medulloblastoma (MB) tumorspheres overexpressing Myc (Trp53–/– overexpressing Myc) and values compared to Trp53–/– controls (n = 10 biologically independent samples). c, d RT-qPCR of (c) MYC, MITF, TFEB, TFE3 and (d) lysosomal genes in U2OS osteosarcoma cells with silenced MYC expression. e RT-qPCR of ARSA, LAMP1, MCOLN1 and NEU1 was performed in HeLa cells silenced for MYC (shMYC), (n = 6 biologically independent samples) and treated with SAHA (8 μM for 24 h). f Activity assays for NEU1 in HeLa cells silenced for MYC (shMYC) or infected with a lentiviral empty vector control (Empty) (n = 3 biologically independent samples) treated with DMSO and SAHA (8 μM for 24 h). All graphs are presented as mean ± SD. Statistical analysis was performed using the Student t-test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

8 NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11568-0 ARTICLE

a HeLa: RT-qPCR b 12.0 DMSO DMSO ** 10.0

8.0 *** ****

6.0 ** **** **** *** ** ***

4.0 **** ** ** Fed Starved

relative to DMSO ** **** **** ** *** ** SAHA SAHA 2.0 * Normalized mRNA expression

0 ULK1 ATG5 VPS18 ATG12 ATG4A ATG4B BECN1 ATG4D ATG4C PIK3R4 PIK3C3 UVRAG AMBRA1 SQSTM1 ATG16L2 GABARAP MAP1LC3A MAP1LC3B GABARAPL1 GABARAPL2 Fed Starved

c deFED Starved 15 **** * ** BAF BAF 64.0 *** * **** DSDS DS D S **** * 2 **** 10 LC3 I 16.0 ** **** ** ** ** LC3 II ** Short 4.0 *

Puncta/mm 5 LC3 I 1.0 LC3 II *** relative to DMSO FED *** 0 Long

Normalzied LC3B protein levels 0.25 DSDSDSDS SAHA SAHA DMSO DMSO BAF BAF FED Starved FED Starved Loading

f Color key g h HeLa: ChIP Acetyl-H3K14

H1ESC HEPG2 K562 0 0.4 0.8 p = 0.0495 Value 1.0 HeLa: ChIP HDAC2 AMBRA1 ATG10 8.0 ATG12 DMSO ATG16L1 0.8 DMSO IgG DMSO ATG16L2 SAHA DMSO IgG ATG2A 6.0 SAHA ATG2B 0.6 SAHA IgG ATG3 SAHA IgG ATG4A ATG4B 4.0 ATG4D 0.4 % of input ATG5 % of input ATG9A ATG9B BACN1 0.2 2.0 BID EEF2 EEF2K EIF2AK2 0 0 EIF2AK3 MAP1LC3B MAP1LC3B EIF2S1 EIF4EBP1 EIF4G1 FKBP1A i j FKBP1B MEFs: RT-qPCR MEFs: RT-qPCR HIF1A p = 0.0031 MAP1LC3B p = 0.002 MAP1LC3C 12.0 3.0 p = 0.004 MAP2K7 MAPK1 MLST8 MTMR14 9.0 PARP1 RGS19 2.0 RHEB RPS6KB1 SIRT1 6.0 SIRT2 SQSTM1 TSC1 1.0 TSC2 relative to DMSO ULK1 3.0 relative to DMSO ULK2 ULK3 USP10 Normalized mRNA expression ATG4C Normalized mRNA expression ATG7 0 0 MAP1LC3A MTOR PARK2 RPTOR TUSC1 UVRAG WT + shC WT + shC WT + shC WT + shC dKO + shC dKO + shC dKO + shC dKO + shC –5 K +5 K –5 K TSS+5 K –5 K TSS +5 K –5 K TSS+5 K –5 K TSS +5 K –5 K TSS +5 K dKO + shMitf dKO + shMitf dKO + shMitf TSS dKO + shMitf Gabarapl2 Map1lc3b Pik3c3 Sqstm1 enzyme in all treated fibroblasts, which was paralleled by MYC antagonizes MiT/TFE in cancer and pluripotent stem enhanced NEU1 residual enzyme activity (Fig. 8b, c and Sup- cells. Collectively our data suggest a rheostat model based on plementary Fig. 7). ChIP performed with acetylated mutual exclusivity between MYC and MiT/TFE expression. We H3K14 antibody showed acetylation of the NEU1 promoter sought to investigate this paradigm in the context of cell pro- region also in sialidosis fibroblasts, confirming NEU1 tran- liferation and differentiation programs, which are known to be scriptional activation (Fig. 8d). These findings potentially associated with high or low levels of MYC27 We chose as broaden the use of HDAC inhibitors for ameliorating the dis- model systems tumor tissue samples and human induced ease phenotype in some forms of LSDs. pluripotent stem cells (hiPSCs) reprogrammed from human

NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11568-0

Fig. 5 HDACs epigenetically regulate autophagy. a RT-qPCR of autophagy genes in SAHA-treated (20 μM for 24 h) HeLa (n = 5 biologically independent samples). b EGFP-LC3 expression in HeLa treated with DMSO or SAHA (20μM for 24 h) under fed or starved conditions (EBSS). Scale bar, 20 μm. c Quantification of EGFP-LC3 (n = 12 images). d Top, representative image of anti-LC3 immunoblot of HeLa treated with DMSO (D) or SAHA (S) (20 μM for 24 h) under fed or starved conditions (EBSS) with or without Bafilomycin A1 (BAF) to assess the autophagic flux; short and long exposure. Bottom, Coomassie stained immunoblot used as loading control e Quantification of LC3II/LC3I ratio (n = 5 independent experiments) normalized to the loading control and relative to DMSO FED values; D = DMSO, S = SAHA, starved = EBSS treatment. f HDAC2 occupancy of autophagy gene promoters analyzed using ChIP-seq datasets from the ENCODE/Haib project. Genes on the left of the vertical black line are occupied by HDAC2. p = 4.656e-08, odds ratio = 6.0, Fisher’s exact test. g ChIP analyses of the MAP1LC3B promoter using anti-HDAC2 antibody was performed in HeLa treated with SAHA (20μM for 24 h) or DMSO (n = 6 biologically independent samples); IgG control (n = 3 biologically independent samples). h ChIP analysis of the promoter of MAP1LC3B using acetyl histone H3 Lys 14 antibody (Acetyl-H3K14) in HeLa treated with SAHA (20 μM for 24 h) or DMSO (n = 6 biologically independent samples); IgG control (n = 3 biologically independent samples). i, j RT-qPCR of (i) Gabarapl2, Map1lc3b and (j) Pik3c3 and Sqstm1 was performed after SAHA treatment (20μM for 24 h) in WT MEFs, in MEFs with CRISPR-mediated knockout of Tfeb and Tfe3 (dKO) and in MEFs dKO silenced for Mitf (dKO + shMitf) (n ≥ 6 biologically independent samples). shC refers to cells transduced with shRNA control lentivirus. Statistical analysis was performed comparing the normalized expression of the autophagy genes in MEFs WT + shC versus dKO + shC and in MEFs WT + shC versus dKO + shMitf. Boxes represent the mean value and bar inside the box represents median value; upper bar represents maximum of distribution; lower bar represents minimum of distribution (95% confidence level). Graphs are presented as mean ± SD. Statistical analysis was performed using the Student t-test. *p < 0.05, **p < 0.01, ****p < 0.0001

fibroblasts, because the processes of carcinogenesis and repro- epigenetically and transcriptionally regulate these cellular gramming share many common features: both cancer cells and catabolic machineries. By modulating the levels of available hiPSCs have a sustained proliferative potential, replicative MYC, HDACs bestow an inhibitory loop on lysosomal and immortality, and lose their original differentiation state50. Fur- autophagy genes. Whether this is because HDACs regulate thermore, upon oncogenic activation, cancer cell progenitors MYC transcription, as our current data seem to suggest, or acquire stem cell-like characteristics by inducing a dediffer- because knockdown of HDACs leads to decreased cell pro- entiation program50. For this purpose, we first assessed the levels liferation and therefore MYC is downregulated, remains a of MYC and MiT/TFE proteins in a human tumor tissue question to be addressed in future studies. In either case, the microarray derived from colon adenocarcinoma, and in patient- effect of reduction of MYC levels is to free the targeted pro- derived xenografts from group 3 medulloblastoma and rhabdo- moter sites of lysosomal and autophagy genes, thereby allowing myosarcoma. The immunohistochemistry (IHC) findings for the binding by MiT/TFE members or FOXH1, and activation of each of the showed that neoplastic cells were hetero- their transcription. This model may explain how lysosomal/ genous in their nuclear and cytoplasmic expression of MYC, autophagy function changes within a cell, and how cells tran- HDAC2 and TFE3. Importantly, the nuclear expression of MYC sition and adapt through different metabolic states. Further- and HDAC2 were common in subpopulations of neoplastic cells, more, our data suggest that this epigenetic rheostat has regardless of their ontogeny. On the other hand, TFE3 appeared important implications for the treatment of LSDs. Thus, to be more frequently localized to the cytoplasm in neoplastic changes in the concentrations of HDACs, MYC, and MiT/TFE cells that expressed MYC and HDAC2 in their nuclei (Fig. 9a, b in LSDs may contribute to the clinical and phenotypic het- and Supplementary Fig. 8a–e). erogeneity of these disorders. In agreement with this hypoth- Reprogrammed hiPSCs that showed a normal karyotype and esis, 11 HDAC genes have been found upregulated in fibroblasts had no viral integrations (Supplementary Fig. 9a, b) were from patients with Niemann-Pick Type C disease12. subjected to comparative proteomic profiling with their parental Until the discovery of MiT/TFE involvement in lysosomal fibroblasts. We found that in hiPSCs the increased levels of both biogenesis, lysosomes were mostly considered the recycling/ MYC and HDAC2 proteins were counterbalanced by a clear degradation center of the cell, and their housekeeping-type genes decrease in the levels of TFEB (Fig. 9c), a finding that explains the were thought to require little or no regulation. MiT/TFE studies reduced expression of both autophagy and lysosomal proteins in demonstrated the existence of a broader network of transcription hiPSCs compared to their parental fibroblasts (Fig. 9d, e). factors, regulating mainly autophagy function. In addition, a large Western blot analyses performed on hiPSCs and parental body of literature has focused on how modifications of histones fibroblasts confirmed the activation of MYC and HDAC2 (Fig. 9f regulate autophagy52,53. However, none of these studies addressed and Supplementary Fig. 9c–f) and consequent decreased levels of the simultaneous regulation of these two catabolic machineries the lysosomal NEU1 and LAMP1 (Fig. 9f and Supplementary that we have now investigated. Our results demonstrate the Fig. 9g–j). Finally, correlative analysis of microarray of hiPSCs at concurrent involvement in these processes of transcription factors different reprogramming stages51 demonstrated that the expres- and nucleus-residing HDACs. Several HDACs are known to sion levels of FOXH1 positively correlated with those of a subset regulate autophagy, but not lysosomal biogenesis. However, their of autophagy genes (Supplementary Table 4). Together, these data effects are cell-type specific and mainly involve cytosolic acet- suggest that HDAC/MYC and MiT/TFE or FOXH1 define ylation of targeted proteins, rather than acetylation of histones in distinct cellular fates and are expressed in a mutually exclusive the nucleus54,55. In support of a nuclear control of autophagy, manner to control lysosomal biogenesis and autophagy. inhibition of HDAC1 has been shown to induce autophagy and to promote in several types of cancer cells56,57. Our Discussion proposed mode of autophagy regulation by antagonistic tran- “ The notion that activation or repression of lysosomal biogenesis scription factors may share similarities to the autophagy master ”58 “ ” 59 and autophagy could be simultaneously controlled by epige- switch , and the hepatic nutrient-sensing models previously netic and transcriptional mechanisms has been hypothesized, proposed. In our model, however, MYC not only opposes the but until now, no evidence has been shown for such coordi- activity of MiT/TFE on lysosomal and autophagy gene promoters, nated regulation. In this study, we uncovered a hierarchical and but also modulates their levels in an acetylation-dependent dynamic rheostat that couples HDACs with MYC to manner.

10 NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11568-0 ARTICLE

a 5 kb 2 kb b 2.0 HeLa: ChIP MYC

60 Myc 90 Myc p = 0.021 1.6 Tags

60 Inp 90 Inp DMSO 1.2 DMSO IgG SAHA Vps18 Atg4d TSS TSS SAHA IgG 2 kb 2 kb Myc Myc 50 250 % of input 0.8 Tags Tags Tags 0.4

50 Inp

250 Inp

0 Map1lc3b Gabarapl2 MAP1LC3B TSS TSS

cdeMB: RT-qPCR ***** **** **** ** MB 1.5 p = 0.0325 **** **** ** *** 1.0 G3c G3b G3d G3e kDa Control G3a 1.0 0.1 20

relative to control 15 LC3 I 0.5 LC3 II 0.01

Normalized mRNA expression 10 ULK1 Quantification of LC3 II levels ATG5 VPS18

ATG12 0 ATG4A ATG4B BECN1 ATG4D ATG4C PIK3R4 PIK3C3 UVRAG AMBRA1 SQSTM1 CG3 ATG16L2 GABARAP MAP1LC3A MAP1LC3B MB GABARAPL1 GABARAPL2

f U2OS (siMYC) cells g 2.5 9.0 **** HeLa shMYC: RT-qPCR **** **** 7.5 2.0 **** **** *** 6.0 **** *** 1.5 *** *** 4.5 *** 1.0 *** 3.0 ** **

relative to siControl ** 0.5 1.5 Normalized mRNA expression Normalized mRNA expression 0 0 ULK1 ATG5 PIK3C3 AMBRA1 ATG16L2 GABARAP Empty SAHA Empty SAHA Empty SAHA Empty SAHA GABARAPL1 Empty DMSO Empty DMSO Empty DMSO Empty DMSO shMYC SAHA shMYC SAHA shMYC SAHA shMYC SAHA shMYC DMSO shMYC DMSO shMYC DMSO shMYC DMSO GABARAPL2 MAP1LC3B PIK3C3 SQSTM1

Fig. 6 MYC represses autophagy. a ChIP-seq analysis for Myc binding to the promoters of Vps18, Map1lc3b, Atg4d, and Gaparapl2 performed in mouse group 3 medulloblastoma tumorspheres overexpressing Myc,MB(Trp53−/− overexpressing Myc)(n = 3 independent experiments). b ChIP analyses of the MAP1LC3B promoter using anti-MYC antibody performed in HeLa cells treated with SAHA (20 μM for 24 h) or DMSO (n = 3 biologically independent samples). c RT-qPCR of autophagy genes in tumorspheres Trp53−/− overexpressing Myc (MB), (n = 10 biologically independent samples). d Comparative LC3 immunoblots of tumorspheres Trp53−/− overexpressing Myc and control Trp53−/− cells. e Quantification of the LC3II/LC3I ratio normalized to Coomassie stained immunoblots, used as loading control. C = Control (Trp53−/−) cells (n = 3 biologically independent samples); G3 = tumorspheres Trp53−/− overexpressing Myc (n = 15 biologically independent samples). f Expression levels of several autophagic genes in U2OS cells with silenced MYC expression. g RT-qPCR of GABARAPL2, MAP1LC3B, PIK3C3 and SQSTM1 was performed in HeLa cells silenced for MYC (shMYC) (n = 3 biologically independent samples) and treated with SAHA (8 μM for 24 h). Boxes represent the mean value and bar inside the box represents median value; upper bar represents maximum of distribution; lower bar represents minimum of distribution (95% confidence level). Graphs are presented as mean ± SD. Statistical analysis was performed using the Student t-test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

The discovery that FOXH1 regulates autophagy increases the are regulated by phosphorylation and shuttle between the cyto- number of transcription factors that control this process, plasm and the nucleus, where they activate or inhibit autophagy although their interplay and response to different intrinsic or genes60,61. FOXO1, one of the most abundant FOXOs, can induce extrinsic cues remains to be determined. FOXH1 belongs to the autophagy in the by directly binding to autophagy-related FOXO family of transcription factors, which are pivotal during proteins62. This transcription factor has also been shown to development and are emerging as key regulators of autophagy. prevent cellular dedifferentiation during early endocrine devel- One member, FOXO3, was the first transcription factor found to opment63. Considering the role of FOXH1 in patterning the control autophagy60. Like the MiT/TFE family members, FOXOs anterior primitive streak and in nodal signaling47,64, the possible

NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11568-0

a b 3.0 HeLa FOXH1: RT-qPCR Motif/name Rank P-value 2.5 ** ** 1 1e–3 2.0 * **** Foxh1(Forkhead)/hESC-FOXH1-ChiP-Seq **** **** (GSE29422)/Homer 1.5 **** *** to DMSO **** 2 1e–2 1.0 Normalized mRNA expression relative 0.5 E-box(HLH)/Promotor/Homer ATG5 ATG4A BECN1 ATG4C 3 1e–2 PIK3C3 ATG16L2

Fli1(ETS)/CD8-FLI-ChiP-Seq(GSE20898)/Homer GABARAP MAP1LC3B GABARAPL2 c e f g kDa NT FOXH1 p < 0.0001 20 p = 0.0003 40 6.0 LC3I kDa DMSO SAHA 15 LC3II 250 30 d p = 0.0331 10.0 4.0 100 75 20 8.0 50

6.0 protein levels expression normalized mRNA 2.0 37 10 4.0

25 Normalized Flag-FOXH1 LC3 II levels 2.0 FOXH1 20 0 Quantification of 0 HeLa DMSO SAHA 0 DMSO SAHA HeLa DMSO SAHA HeLa Flag-FOXH1 FOXH1

h ijp < 0.0001 k p = 0.0059 150 2.0 200 U2OS cells 0.5 kb p = 0.0021 15

y Inp Myc 1.0 150 100 0.5 Tags 0.25 100 normalized 15 0.12 normalized 50

normalized mRNA 0.06 50 Foxh1 mRNA expression mRNA expression FOXH1 0.03 Tags

FOXH1 0 Foxh1 0 expression relative to control TSS Control DoxMYC CG3 siCntr siMYC U2OS MB U2OS

Fig. 7 FOXH1 is an additional activator of autophagy and responds to the HDACs/MYC axis. a Motif analysis using HDAC2-binding sites for autophagy genes identified the FOXH1 motif. b RT-qPCR of autophagy genes in FOXH1-overexpressing HeLa cells (n = 6 biologically independent samples) p < 0.01. c Representative LC3 immunoblot from FOXH1-overexpressing cells. d Quantification of the LC3II/LC3I ratio normalized to Coomassie stained immunoblots, used as loading control (n = 6 independent experiments). e FOXH1 RT-qPCR in SAHA-treated HeLa cells (20 μM for 24 h; n = 7 biologically independent samples). f Representative immunoblot of FOXH1 overexpression in HeLa cells treated with DMSO or SAHA (20 μM for 24 h). g Quantification of f (n = 3 independent experiments) normalized to Coomassie stained immunoblots, used as loading control. h MYC binding to the FOXH1 obtained from ChIP-seq datasets for U2OS cells overexpressing MYC. i Expression of the FOXH1 gene in U2OS cells with MYC overexpression, data obtained from Walz et al. j RT-qPCR of Foxh1 in tumorpheresTrp53−/− overexpressing Myc. Values are relative to those obtained in control Trp53−/− cells (n = 5 biologically independent samples). k RT-qPCR of FOXH1 in U2OS cells in which MYC was silenced by shRNA. Boxes represent the mean value and bar inside the box represents median value; upper bar represents maximum of distribution; lower bar represents minimum of distribution (95% confidence level). Graphs are presented as mean ± SD. Statistical analysis was performed using the Student t-test involvement of this transcription factor in embryonic specifica- correlates with the presence of acetylated, transcriptionally- tion via regulation of autophagy merits future investigation. permissive chromatin, our results suggest a stage-specific We also uncovered a previously uncharacterized antagonistic requirement for HDAC2 in reprogramming, in line with a role between MYC and MiT/TFE that controls gene expression recent publication demonstrating that HDAC2 inhibits the early programs during cell differentiation. The proposed mechanism phase of reprogramming, but positively affects the final stages of implies that the acquisition of “stemness” requires robust acti- this process65. vation of MYC and the active repression of lysosomal and Cellular reprogramming shows remarkable parallels with the autophagy genes. In contrast, the maintenance of a fully differ- formation of cancer stem cells50. The factors necessary for entiated state requires activation of lysosomal and autophagy somatic reprogramming are also oncogenic drivers and promote signatures by MiT/TFE members. In agreement with this the re-acquisition of early developmental stages and unlimited hypothesis, we observed upregulation of MYC and down- proliferation66. MYC, for instance, is required to potently gen- regulation of TFEB in hiPSCs when compared to fully differ- erate fully reprogrammed hiPSCs67–69. It is also overexpressed in entiated isogenic fibroblasts. The mutually exclusive expression of more than 70% of human tumors, driving tumorigenesis and these factors, likely regulated by HDAC2, ensures repression of maintenance70. Blocking MYC activity causes tumor regression the lysosomal and autophagic systems in hiPSCs. Although it has by promoting cell cycle arrest and differentiation71. Our studies been amply demonstrated that induction of pluripotency support the notion that tumors may use the HDAC/MYC–MiT/

12 NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11568-0 ARTICLE

a Human fibroblasts: RT-qPCR c 64.0 12.0 * **** 32.0 *** **** 8.0 * 16.0 * 4.0 8.0 **** 4.0 2.0 2.0 1.0 1.0 NEU1 activity (nmol/mg/h) Normalized mRNA expression 0.5 0.5 DSDSDSDS DSDSDSD S Type I Type II Type I Type II Type I Type II Type I Type II

b ControlType I Type II Type I Type II d DS DSDS D SDS Human fibroblasts: kDa ChIP Acetyl-H3K14 75 0.10 p = 0.0139 DMSO 50 DMSO IgG 0.08 SAHA SAHA IgG 37 0.06 Anti-NEU1

0.04 75 % of input

0.02 50

37 0 Coomassie NEU1

Fig. 8 HDAC inhibition in LSD. a RT-qPCR of NEU1 in SAHA- or DMSO-treated sialidosis fibroblasts, (SAHA 20 μM for 24 h; n = 8 biologically independent samples). Type I = attenuated form of sialidosis; Type II = severe form of sialidosis. b Top, representative immunoblot of fibroblasts isolated from two patients with sialidosis type I and two patients with sialidosis type II were treated with SAHA (S) (20 μM for 24 h) or DMSO (D) and probed with anti- NEU1 antibody; bottom, Coomassie stained immunoblot used as loading control. c NEU1 activity in fibroblasts isolated from two patients with sialidosis type I and two patients with sialidosis type II treated with SAHA (S)- or DMSO (D)-treated (SAHA, 20 μM for 24 h; n = 7 biologically independent samples). d ChIP analysis of the promoter of NEU1 performed with acetyl histone H3 Lys 14 antibody (Acetyl-H3K14) in fibroblasts from one type I sialidosis fibroblasts treated with SAHA (20 μM for 24 h) or DMSO (n = 3 biologically independent samples). Boxes represent the mean value and bar inside the box represents median value; upper bar represents maximum of distribution; lower bar represents minimum of distribution (95% confidence level). Graphs are presented as mean ± SD. Statistical analysis was performed using the Student t-test. *p < 0.05, ***p < 0.001, ****p < 0.0001

TFE axis to repress their lysosomal and autophagy systems, a serum, 2 mM GlutaMAX, penicillin (100 U/mL), and streptomycin (100 μg/mL) fi μ paradigm that may be exploited therapeutically. In fact, HDAC (Gibco) in a humidi ed 5% CO2 atmosphere with 2 g/mL Puromycin. Skin fibroblasts from control individuals were obtained from Coriell Institutes; inhibitors have recently shown therapeutic potential in group 3 fi 72 human sialidosis broblasts type I or type II received by my laboratory were medulloblastoma, as a single-agent therapy or in combination uncoded and unidentifiable. They came from the Rotterdam Biobank (Rotterdam, with PI3K inhibitors73. Given that HDAC inhibitors increase The Netherlands), the Pediatric Undiagnosed Diseases Program, National Human FOXO1 expression, thereby inhibiting tumor growth, and that Genome Research Institute/NIH (Bethesda MD, USA) and the Muscle Unit Section silencing FOXO1 dampens this effect, we are tempted to speculate and Laboratory of Skeletal Muscle Pathology Department of Neurology, Medical School of the University of São Paulo (São Paulo, Brazil). Original consent was that autophagy and lysosomal biogenesis activated by MiT/TFE obtained by the clinicians from the patient or a family member and the study was and/or FOXH1 may be the pathways of choice for targeted cancer approved by the ethics committees of the three institutions. All cells were banked therapy. Future studies will reveal the generality of our rheostat for secondary future research. model in different cancer types, and how it contributes to tumor For transient expression, HeLa cells were plated on 4-chambered Lab-Tek glass coverslips (Nalge Nunc Int.) and transfected with the FOXH1 plasmid74 and with initiation, progression and maintenance. the EGFP-LC3 plasmid using the calcium phosphate precipitation method. Transfected cells were washed with PBS 3 times and fixed for 10 min in PBS containing 4% PFA and then washed extensively after fixation. Slides were fi Methods mounted with ProLong Antifade Mountants (Thermo sher) and imaged using a confocal fluorescence microscope (Zeiss LSM 780). LC3 positive autophagosome Cell culture and treatment. Human cell lines HeLa and RH30 are available at the were counted with the Zeiss software in more than 10 fields; ROI(s) or “objects” ATCC and were provided by the laboratory of Dr. Grosveld. Cells were grown at were determined for each image. The number of objects and the area of the ROI 37 °C in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% were recorded and calculated as puncta/mm2. Autophagic flux was assayed probing cosmic calf serum (HyClone), 2 mM GlutaMAX, penicillin (100 U/mL), and western blots with anti-LC3 antibody in HeLa cells treated with bafilomycin A1 streptomycin (100 μg/mL) (Gibco) in a humidified 5% CO atmosphere. Neuro- 2 (Sigma) under FED (complete medium) or Starved conditions. In the latter case, blastoma cells (Sy5y), were grown in a 1:1 mixture of F12 (Gibco) and Eagle’s HeLa cells were washed extensively with a Hanks’ balanced salt solution (HBSS) minimal essential medium (EMEM) (Lonza), both supplemented with 10% fetal (Invitrogen) and incubated for 3–6 h at 37 °C in Earle’s balanced salt solution bovine serum, GlutaMAX, and antibiotics, as indicated for HeLa cells. Tumor- (EBSS) (Sigma-Aldrich). HeLa cells were transfected with two short hairpin RNA spheres from mouse group 3 medulloblastoma (Trp53–/– overexpressing Myc) (shRNA) constructs for human HDAC2 and MYC genes cloned into the pLKO1 (#19251; #19554, #15486; #13465; #19568)40,44, and tumorspheres from Trp53–/– HIV-based lentiviral vector (Dharmacon). MEFs with CRISPR-mediated ablation (used as control) were grown on ultralow-attachment plates and supplemented of Tfeb and Tfe3 were transduced with the GIPZ microRNA-adapted shRNA for with human recombinant basic FGF and EGF (Peprotech) every 3 days. MEFs WT Mitf (Dharmacon). and MEFs with CRISPR-mediated ablation of Tfeb and Tfe3 were grown in Dul- For lentivirus preparation, 293T cells were co-transfected of with 10 μgof becco’s modified Eagle’s medium (DMEM) supplemented with 10% primary pCAGkGP1R and 2 μg of pCAG4RTR2 packaging plasmids, 2 μg of pCAG-VSVG

NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications 13 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11568-0

abAdenocarcinoma; Grade 2 150 Stage 3b Stage 3c p < 0.0001 100

50 positive cells

MYC Percentage of MYC- 0 Nucleus Cytoplasm

p = 0.0002 50

30

TFE3 positive cells 10 Percentage of TFE3- Percentage of 0 Nucleus Cytoplasm

20 p = 0.224

15

10 HDAC2

positive cells 5

Percentage of HDAC2- 0 Nucleus Cytoplasm c d 22.0 Proteomics Human fibroblasts 22.0 Proteomics Human fibroblasts hiPSC hiPSC 20.0 **** 20.0 ** **** **** **** 18.0 18.0 ****

**** **** 16.0 16.0 *** Protein levels Log2 intensity Log2 levels Protein Protein levels Log2 intensity Log2 levels Protein 14.0 **** 14.0 MYC ULK1 TFEB VPS18 HDAC1 HDAC2 SQTM1 AMBRA1 MAP1LC3B MAP1LC3A GABARAPL2 GABARAPL1

efHuman fibroblasts 22.0 Proteomics hiPSC Fibro iPSC Fibro iPSC **** 75 50 20.0 **** **** 37 **** 50 **** *** **** Anti-HDAC2 Anti-NEU1

18.0 *** ****

16.0 Fibro iPSC Fibro iPSC **** 75 Protein levels Log2 intensity Log2 levels Protein 14.0 100 50 75 TPP GLA GBA GNS GLB1 NEU1 NPC1 CTSA CTSD ARSA

LAMP1 Anti-MYC Anti-LAMP1 envelope plasmid, and 10 μg of vector plasmid (shHDAC2, shMYC), or 7 μgof diluted in DMSO (Sigma-Aldrich). For romidepsin treatment, cells were incubated psPAX2 (a gift from Didier Trono, Addgene plasmid # 12260; http://n2t.net/ for 24 h at 37 °C in medium containing 10 nM romidepsin (Active Motif) diluted in addgene:12260; RRID:Addgene_12260), 2 μg pCAG-VSVG envelope plasmid and DMSO (Sigma-Aldrich). Medium containing DMSO, without SAHA or 10 μg of vector plasmid (shMitf, Dharmacon). An empty vector or an shRNA romidepsin, was used as control. control vector were used as controls. Stable cells were selected with puromycin (2 μg/mL, Sigma- Aldrich) or in the case of GIPZ with FACS. Cells were analyzed 72 h after transfection. For SAHA treatment, cells were incubated for 24 h at 37 °C Western blot analysis and immunoprecipitation. Cells were lysed in RIPA buffer in medium containing 20 μM or when specified 8 μM SAHA (Sigma-Aldrich) [0.1% sodium dodecyl sulfate (SDS), 1% sodium deoxycholate, 1% Triton X-100,

14 NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11568-0 ARTICLE

Fig. 9 HDACs/MYC–MiT/TFE axis in cancer and pluripotent stem cells. a Immunohistochemical analysis of MYC, TFE3, and HDAC2 in colon carcinoma samples. Regions marked by thick black arrows show that the nuclear expression of MYC and HDAC2 are common in subpopulations of neoplastic adenocarcinoma cells, especially at the invasive edges of the tumors. TFE3 appears to be more frequently localized to the cytoplasm in these same cells. Scale bar 25 μm. b MYC, TFE3, and HDAC2 expression levels in the cytoplasm and the nucleus of neoplastic cells were scored separately (n = 6 independent tumors). c–e Protein levels of (c) HDAC1, HDAC2, TFEB, and MYC, (d) of lysosomal enzymes and lysosomal membrane components and (e) of autophagy effectors detected in proteomics of human fibroblasts (n = 4 biologically independent samples) and reprogrammed hiPSCs, (n = 7 biologically independent samples). f Representative immunoblots of hiPSCs and their parental fibroblasts probed with anti-MYC, anti-HDAC2, anti-NEU1 and anti-LAMP1 antibodies. Boxes represent the mean value and bar inside the box represents median value; upper bar represents maximum of distribution; lower bar represents minimum of distribution (95% confidence level). Graphs are shown as mean ± SD. Statistical analysis was performed using the Student t-test. **p < 0.01, ***p < 0.001, ****p < 0.0001

140 mM NaCl, 10 mM Tris, inhibitors cocktail, and phosphatase inhibi- Microarray analysis. Total RNA (100 ng) extracted from DMSO- and SAHA- tors]. Protein concentration was determined by optical density (A595) using the treated HeLa cells (three biological replicates of each condition) was converted into Pierce bovine serum albumin (BSA) protein assay (Pierce). For immunoprecipi- biotin-labeled cRNA (Ambion WT Expression Kit, Affymetrix Inc) and hybridized tation, HeLa cells were lysed in RIPA buffer with protease inhibitors cocktail, and to a Human Gene 2.0 ST GeneChip (Affymetrix Inc). Probe signals from arrays μ phosphatase inhibitors (Roche). Total protein lysates (250 g) were incubated were normalized and transformed into log2 transcript expression values using the overnight at 4 °C on a rotating platform with 2 μg of anti-MYC antibody (Cell Robust Multiarray Average algorithm (Partek Genomics Suite v6.6). GSEA (http:// Siganling) or normal immunoglobulin G (IgG) (Cell Signaling). Dynabeads Protein software.broadinstitute.org/gsea/index.jsp) was performed using curated pathways G (Millipore) were simultaneously incubated with V5 tag (1 μg/ml) obtained from MolSigDB (Broad Institute). Differentially expressed transcripts antibody–blocking peptide. Beads were washed extensively with RIPA buffer and were identified by ANOVA and the false discovery rate (FDR) was estimated. added to the cell lysates; samples were incubated for an additional hour. The beads Functional enrichment analysis of gene lists was performed using the DAVID were then washed four times with RIPA buffer. Proteins were eluted with sample bioinformatics databases (http://david.abcc.ncifcrf.gov/). loading buffer before immunoblotting. Immunoprecipitates were run on 4-20% Bis-Tris mini protein gels (Biorad) and transferred ON. Membranes were immu- noblotted for anti-HDAC2 antibody (Cell Signaling #2545, 2μg) and anti-MYC Immunohistochemistry. Immunohistochemistry (IHC) analyses were performed μ fi antibody (Cell Signaling 9402, 2μg). For immunoblot analyses, HeLa, RH30, Sy5y on 10- m-thick paraf n sections obtained from one RH30R xenograft, one and medulloblastoma cells were lysed in RIPA buffer; MEFs with CRISPR- medulloblastoma subgroup 3 xenograft and from a tissue microarray containing 6 mediated ablation of Tfeb and Tfe3 were lysed in 25 mM Hepes‐KOH, pH 7.4, cases of colon adenocarcinoma (TMA CO243b, US Biomax, Inc.). After the 150 mM NaCl, 5 mM EDTA, and 1% Triton X‐100 (w/v) supplemented with blocking solution (0.1% BSA, 0.5% Tween-20, and 10% normal serum) step, sec- fi protease and phosphatase inhibitors (Roche). MEF lysates were incubated on ice tions were incubated overnight at room temperature with the speci c antibody for 30 min, mechanically sheared through a 25‐gauge needle. and centrifuged at diluted in blocking buffer (anti-MYC, Cell Signaling 9402, 1:300, anti-TFE3 Sigma 16,000 × g for 10 min at 4 °C. Protein lysates (20-35 µg) underwent electrophoresis HPA023881 1:300 and anti-HDAC2, Abcam ab7029 1:500). The sections were on NuPage Novex (4–12%), 10, 12, and 4–20% Bis-Tris mini protein gels (Life washed and incubated with biotinylated secondary antibody (Jackson ImmunoR- Technologies or Biorad) and were wet-blotted onto PVDF membranes. Membranes esearch Laboratory) for 1 h. Endogenous peroxidase was removed by incubating were blocked in Tris-buffered saline containing 0.1% Tween-20 (TBS-T) and 5% the sections with 0.1% hydrogen peroxidase in PBS for 30 min. Antibody detection nonfat dry milk and incubated overnight at 4 °C with primary antibodies in either was performed using the ABC Kit and diaminobenzidine substrate (Vector 3% BSA in TBS-T solution or 5% milk in TBS-T. The following day, membranes Laboratories) and sections were counterstained with haematoxylin according to were incubated with HRP-conjugated or fluorescent-tagged secondary antibodies standard method. The stained slides were scanned with an Aperio ScanScope XT and developed using SuperSignal West Femto Maximum Sensitivity Substrate scanner (Leica Biosystems) to create whole slide images that were scalable up to a fi (Thermo Scientific). The following antibodies were used for all the immunoblot magni cation of ×200. Protein marker expression was scored separately for the analyses: anti-MYC (Cell Signaling 9402, 1:1000), anti-Lamp1 (BD 553792, 1:500), cytoplasm and the nucleus (Aperio ImageScope software, Leica Biosystems). A anti-LAMP1 (Cell Signaling 9091, 1:1000), anti-LC3B (Cell Signaling 3868, 1:1000), manual review of all IHC slides was also performed using a Nikon Eclipse Ni anti-Flag M2 (Sigma F1804 1:500), anti-HDAC2 (Abcam 7029, 1:1000, Cell Sig- microscope and correlated with the algorithm generated data. Findings were naling 2545, 1: 1000), anti-TFEB (Bethyl laboratories Inc A303-672A, 1:2000) and reported as the percentage of subpopulations of neoplastic cells per subcellular anti-TFE3 (Sigma-Aldrich HPA023881, 1:1000) Anti-NEU1 and anti-PPCA anti- location based on a combination of the automated and manual assessments. bodies were generated in-house. Uncropped and unprocessed scans of the immunoblots shown in Fig. 2d, e are Chromatin immunoprecipitation. Each chromatin immunoprecipitation (ChIP) included in Supplementary Fig. 4c, d; those shown in Fig. 5d are included in analysis was performed with 5 × 106 cells by following a published procedure75. Supplementary Fig. 5l; those shown in Fig. 7c are included in Supplementary Cells were cross-linked with 1% formaldehyde for 10 min at room temperature, Fig. 6c. and the reaction was stopped by adding glycine (final 125 mM) for 5 min. The cross-linked cells were lysed in 800 μL buffer (50 mM Tris pH 8.1, 10 mM EDTA, 1% SDS), supplemented with a protease inhibitor cocktail (Roche). After dilution in Enzymatic activities. NEU1, β-hexosaminidase, α-mannosidase and β-galactosi- 2.0 mL IP buffer (25 mM Tris pH 8.0, 150 mM NaCl, 0.01% SDS, 0.5% deox- dase activities were assayed using the synthetic substrates, 2′-(4-methylumbelli- ycholate, 1% TritonX-100, and 5 mM EDTA), cell lysates were sonicated to gen- feryl)-α-D-N-acetylneuraminic acid; sodium salt, 4MU-N-acetyl- β-D- erate DNA fragments of 200- to 600-bp. Fragmented chromatin was glucosaminide, 4MU-α-D-mannopyranoside, and 4 Methylumbelliferryl β-d- immunoprecipitated with salmon sperm DNA/protein A agarose beads (Millipore) Galactopyranoside, respectively. Briefly, cells were collected and lysed in water. with specific antibodies. The antibodies used for the ChIP experiments were: anti- Lysates were incubated with the substrate in triplicate in 96-well plates for 1 h at HDAC2 (Abcam ab7029, 5 µg), anti-MYC (Cell Signaling 9402, 1:50), anti-TFEB 37 °C. To stop the enzymatic reaction, 200 μL 0.5 M carbonate buffer, pH 10.7, was (Cell Signaling 4240, 1:200), anti-TFE3 (Sigma HPA023881, 5 µg), anti-Acetylated added to all wells. The fluorescence was measured on a plate reader (EX-355, EM- Histone 3 K14 (Cell Signaling 7627 1:50). Agarose beads were also incubated with 460). The net fluorescence values were compared with those of the linear 4MU normal IgG (Cell Signaling 3900). The pulled-down immune-complexes were standard curve and were used to calculate the specific enzyme activities. Activities washed twice each with low-salt buffer (50 mM Tris pH 8.0, 150 mM NaCl, 0.1% were calculated as nanomoles of substrate converted per hour per milligram of SDS, 0.5% deoxycholate, 1% NP40, 1 mM EDTA), high-salt buffer (50 mM Tris pH protein (nmol/mg/h). 8.0, 500 mM NaCl, 0.1% SDS, 0.5% deoxycholate, 1% NP40, 1 mM EDTA), LiCl For cathepsin A activity, the synthetic substrate N-carbobenzoxy-L- wash buffer (50 mM Tris pH 8.0, 250 mM LiCl, 0.5% deoxycholate, 1% NP40, phenylalanyl-L-alanine was used. Briefly, cells were collected and lysed in water 1 mM EDTA), and Tris-EDTA buffer (10 mM Tris pH 8.0, 1 mM EDTA) for and cell homogenates (10 µL) were incubated with 100 µl MES Buffer for 1 h at 10 min at 4 °C. Immune-complexes were eluted in elution buffer (1% SDS, 0.1 M 37 °C in the absence or presence of 1.5 mM N-blocked dipeptide Z-Phe-Ala, NaHCO3) 2 times for 15 min and reverse cross-linked at 65 °C overnight with 5 M (10 µL). The reaction was stopped by heating the plate at 100 °C for 5–10 min NaCl (Sigma-Aldrich). The nest day, the eluates were treated with 0.5 M EDTA following cooling on ice for 5 min. 10 µL from each well was transferred to a new (Millipore, 20-158), 1 M Tris-HCl, pH 6.5 (Millipore, 20-160) and 10 mg/mL 96-well plate and 250 µL Buffered Reagent (o-phtaldialdehyde/2-mercaptoethanol/ Proteinase K (Sigma-Aldrich) and incubated for one hour at 45 °C. Reverse cross- ethanol solution in 50 mM Na-Carbonate buffer pH9.5) was added to each well and linked DNA was purified by QIAquick PCR purification kit (Qiagen). ChIPs were incubated for 5 min before measuring the fluorescence (Ex-355, Em-460). Values performed at least 3 times for each of the antibodies. For sequential ChIP assays, were calculated as subtraction between values obtained in presence of substrate chromatin preparations were first incubated with anti-MYC antibody (Cell Sig- minus values attained without substrate and expressed as nmol/mg/min. naling 9402, 1:50); then chromatin immune-complexes were eluted with 10 mM DTT at 37 °C for 30 min, followed by the second ChIP with antibody for HDAC2

NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications 15 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11568-0

(Abcam ab7029, 5 µg). Eluted DNA was then purified and analyzed with standard and female (two non-viral integrated clones) wild type fibroblast cells were gen- RT-qPCR procedure. RT-qPCR of immunoprecipitated chromatin was carried out erated for this study. All the hiPSCs used in this study showed non-viral integration on a CFX96 real-time PCR Machine (BioRad) using RT2 Sybr Green qPCR and a normal diploid karyotype. Mastermix, 2 µL of the immunoprecipitated material or the purified input were used in the reactions with oligonucleotides available from SA Bioscience (Qiagen). Proteomic analysis of human fibroblasts and hiPSCs. BJ and PCS201 fibroblast Oligos designed to cover a region at + 5 Kb from the promoter of NEU1 (NEU1 cells, grown in DMEM + 10% FBS media, were dissociated by trypsin and collected INT) were used as negative control to amplify a region for no antibody binding. by centrifugation. Human iPSCs (originated from BJ and PCS201 fibroblast cells) The values shown in the graphs are expressed as a percentage of the total were grown in mTeSR media (STEMCELL Technologies) under GeltrexTM LDEV- input DNA. Free hESC-qualified reduced growth factor basement membrane matrix (Thermo Fisher Scientific). Cells were dissociated by accutase (Thermo Fisher Scientific) and Quantitative real-time PCR. Total RNA was isolated from DMSO- and SAHA- collected by centrifugation. Cells were lysed and their protein content digested into treated HeLa, RH30, and Sy5y cell lines, romidepsin-treated HeLa and primary peptides by trypsin. After desalting, the peptides were labeled with TMT reagents. fibroblasts from healthy individuals, sialidosis fibroblasts, as well as from Flag- The labeled samples were equally mixed and further fractionated by neutral pH FOXH1–transfected HeLa cells by using the PureLink RNA kit (Life Technologies) reverse phase liquid chromatography. In general, 10 fractions were collected and following the manufacturer’s protocol. DNA contaminants were removed on a further analyzed by low pH reverse phase LC-MS/MS. During ion fragmentation, DNAse I column (Life Technologies), according to the manufacturer’s protocol. the TMT regents are cleaved to produce reporter ions for quantification. The RNA quantity and purity were measured using a Nanodrop Lite spectrophotometer collected data were searched against a database to identify peptides. While the (Thermo Scientific). Complementary DNA was produced using 2 μg total RNA peptides were identified by MS/MS, the quantification was achieved by the frag- with RT2 First Strand Kit (Qiagen; 330401). RT-qPCR was performed using the mented reporter ions in the same MS/MS scans. The peptide quantification data RT2 Sybr Green qPCR Mastermix, 1 μL (50 ng) cDNA, 10 μM primer, and RNAse- were then corrected for mixing errors and summarized to derive protein quanti- free water in a 25-μL reaction volume on a CFX96 real-time PCR machine fication results. Statistical analysis was performed to determine cutoff for altered (BioRad). The oligonucleotides used for RT-qPCR were purchased from Biorad. proteins and to evaluate associated false discovery rate. Samples were normalized by using HRTP1 (human) or 18S ribosomal RNA (mouse) expression. The plotted values represent the relative normalized expres- HDAC2 binding. The HDAC2 occupation data were downloaded from the Ency- sion of the mRNA in SAHA-treated and in romidepsin-treated cells. DMSO- clopedia of DNA Elements (ENCODE) working group using the Jan 2011 data freeze treated cells were used as control for normalization of the expression for SAHA- (ftp://ftp.ebi.ac.uk/pub/databases/ensembl/encode/integration_data_jan2011). The and romidepsin-treated cells. RT-qPCR values obtained in tumorspheres from optimal unified peak lists called with PeakSeq from two biological replicates from mouse group 3 medulloblastoma (Trp53–/– overexpressing Myc) (#19251; #19554, h1ESC, K562 leukemia and HEPG2 liver carcinoma cells present in Haib data sets were #15486; #13465; #19568) are shown as relative to gene expression values obtained used (http://genome.ucsc.edu/cgi-bin/hgTrackUi?db=hg19&g=wgEncodeHaibTfbs). from tumorspheres from Trp53–/–. Agenewasdefined as bound by HDAC2 if a peak summit was within 5 kb of the transcript start sites. Fisher’s exact test was calculated for this analysis. Lysosomal volume. The lysosomal volume was analyzed as previously described76. Lysotracker green DND 26 (200 nM, Molecular Probes) was added to cells for Motif analysis. HDAC2-binding sites for all lysosomal genes or autophagy genes 40 min. After incubation, cells were washed with PBS, counted, and analyzed by from all three cell lines downloaded from the ENCODE/Haib data set (http:// fluorescence-activated cell sorting (FACS). Mean intensities of lysotracker green = = fl genome.ucsc.edu/cgi-bin/hgTrackUi?db hg19&g wgEncodeHaibTfbs) were uorescence were calculated and plotted. combined. Motif search was performed using a 300-bp sequence near each peak summit, using the program HOMER. Transmission electron microscopy and counting. Transmission electron microscopy analyses were performed at the Electron Microscopy Division of the ChiP and gene expression analyses of available datasets. For ChIP-seq dataset, ’ Cell and Tissue Imaging Center at St. Jude Children s Research Hospital. HeLa cells peak results from MACS2 for U2OS MYC ChIP-seq data (GSE64425)29 were fi treated with DMSO or SAHA were xed in 2.5% glutaraldehyde in 0.1 M sodium associated with transcription start sites (TSS) of genes from the hg19 build of the fi cacodylate buffer. The samples were post- xed in 2% osmium tetroxide and . Transcripts with peaks 5 kb up or down stream of the TSS were dehydrated via a graded series of alcohol, cleared in propylene oxide, embedded in designated as MYC-bound in this cell line. Lists of genes associated with autophagy epon araldite and polymerized overnight at 70 °C. The unstained sections were and lysosomes were then tested for enrichment among bound sites by Fisher’s exact imaged on a JEOL 1200 EX Transmission Electron Microscope with an AMT T2K test. Peak results for medulloblastoma cells ChIP-seq data were obtained from digital camera. More than 20 images were taken from two biological replicates and GSE6442540. Data were remapped to hg19 with bwa/0.7.12, and coverage files were counted blindly for the number of lysosomes. imported into the integrated genome viewer (IGV Broad, MIT), thereby doc- umenting the binding of MYC to lysosomal genes and MiT/TFE members. Peak Reprogramming of fibroblasts into hiPSCs. Human fibroblast (BJ and PCS201 calling of ChIP-seq datasets performed in mouse group 3 medulloblastoma which were male and female wild type, respectively) cells were purchased from tumorspheres lacking Trp53 and overexpressing Myc (control: input samples) was ATCC and were grown in DMEM + 10% FBS media. Once cells were confluent, performed with MACS v.1.4.2 (the keep-dup parameter was adjusted, depending 40 reprogramming was performed by introducing sendai viral vectors (CytoTune 2.0, on the ChIP enrichment at the highest peaks) . For the public RNAseq and fi Invitrogen) expressing four human reprogramming factors (Oct3/4, , , and microarray analysis, microarray raw data les (.Cel) from breast cancer cells c-Myc). In brief, ~5 × 105 fibroblast cells were transduced by Sendai virus. Each (GSE60124), BJ cells (GSE43010) and the count matrix for RNAseq from normal ESC-like colony was picked-up manually by glass pipette under the microscope and endothelial cells (GSE77108) were downloaded from the GEO data base (NCBI). maintained on feeder-free Geltrex coated plates to establish hiPSC lines. After at Array data were rma summarized and log2FC were computed using Partek least five passages of expansion, candidate clones for hiPSC lines were characterized Genomics Suite 6.6 (St Louis, MO USA). For the RNAseq, log2cpm were computed + by expression of pluripotency marker (Oct4, Santa Cruz Biotechnologies, sc-5279, from the matrix using the log(cpm 0.5)/log(2) transformation and these values clone No. C-10, 1:500; Nanog, Santa Cruz Biotechnologies, sc-33759, 1:500; SSEA4, were used to calculate the logFC. Next, a list of lysosomal associated genes (from Millipore, MAB4304, 1:500; SSEA3, Millipore, MAB4303-I, 1:500; TRA-1-60 array annotation, Affymetrix, Mountainview, CA) were compiled, de-duplicated Millipore, MAB4360, 1:500) using an immunofluorescence assay, followed by and matched to the data by gene symbol. For each dataset the logFC of the HDAC karyotype analysis. For the immunofluorescence assay, each hiPSC line was fixed inhibitor vs DMSO control for the logFC for genes in the lysosomal list were immediately with fixing solution (4% paraformaldehyde) and then treated with statistically compared to the remaining logFC values (one tailed t test) and plotted permeabilization buffer (0.2% Triton X-100) after washing with PBS. Cells were in CDF plots using STATA MP/14.1 (College Station, TX USA). For analyses of washed with PBS three times and incubated with blocking solution which con- gene expression on U2OS overexpressing MYC and silenced for MYC, we queried 29 tained 3% BSA in PBS for 15 min. Cells were incubated with anti-Oct4, anti-SSEA4 microarray data from Walz et al. . and anti-SSE3A antibodies in blocking solution overnight. Cells were washed with fl PBS three times and incubated with secondary antibodies (Alexa uor 488 and FOXH1 correlation analysis. Microarray data from GSE5020651 were used for the fl Alexa uor 568, Molecular Probes). After washing with PBS, cells were stained with correlation between autophagy genes and FOXH1. Spearman correlations were fl DAPI to stain nucleus. Each image was examined using a uorescent microscope used to correlate the FOXH1 expression with the autophagy genes present in fl (EVOS, Invitrogen) or by ow cytometry. To exclude the integration of sendai this set. virus, we performed qRT-PCR analysis with a specific probe and primers for a sendai viral gene (Sendai virus detection, SeV; TaqMan probe Mr04269880_mr; Sendai virus detection, SeV/Klf4; TaqMan probe Mr04421256_mr; GAPDH, VIC/ Statistical analyses. Statistical analyses were performed using GraphPad Prism. TAMRA probe 4310884E, Invitrogen). For qRT-PCR analysis, total RNA was Quantitative data are presented as mean ± SD. For comparisons between two extracted from each hiPSC line by using Direct-zol RNA purification Kit (Zymo groups, Student’s t-test (unpaired, two-tailed) was performed. Groups were con- research) and cDNA was synthesized with the ThermoScriptTM RT- PCR system sidered different when p < 0.05. For all quantifications, a minimum of three (Invitrogen). Four hiPSC lines from human male (two non-viral integrated clones) independent experiments were performed. Measurements were taken from distinct samples. Number of replicates is noted in the figure legends.

16 NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11568-0 ARTICLE

Reporting summary. Further information on research design is available in 24. Settembre, C. et al. TFEB links autophagy to lysosomal biogenesis. Science 332, the Nature Research Reporting Summary linked to this article. 1429–1433 (2011). 25. Halazonetis, T. D. & Kandil, A. N. Determination of the c-MYC DNA-binding – Data availability site. Proc. Natl Acad. Sci. USA 88, 6162 6166 (1991). 26. Carroll, P. A., Freie, B. W., Mathsyaraja, H. & Eisenman, R. N. The MYC Expression microarray data are available at the Gene Expression Omnibus repository transcription factor network: balancing metabolism, proliferation and under accession number GSE106175. The data that support the findings of this study are oncogenesis. Front. Med. 12, 412–425 (2018). available from the corresponding author, Alessandra d’Azzo, [email protected], 27. Dang, C. V. MYC, metabolism, cell growth, and tumorigenesis. Cold Spring upon request. Harb Perspect Med 3, https://doi.org/10.1101/cshperspect.a014217 (2013). 28. Li, H. & Wu, X. Histone deacetylase inhibitor, Trichostatin A, activates Received: 18 June 2018 Accepted: 17 July 2019 p21WAF1/CIP1 expression through downregulation of c-myc and release of the repression of c-myc from the promoter in human cervical cancer cells. Biochem. Biophys. Res. Commun. 324, 860–867 (2004). 29. Walz, S. et al. Activation and repression by oncogenic MYC shape tumour- specific gene expression profiles. Nature 511, 483–487 (2014). References 30. Marks, P. A. Discovery and development of SAHA as an anticancer agent. 1. Jaenisch, R. & Bird, A. Epigenetic regulation of gene expression: how the Oncogene 26, 1351–1356 (2007). genome integrates intrinsic and environmental signals. Nat. Genet. 33(Suppl), 31. Telles, E. & Seto, E. Modulation of cell cycle regulators by HDACs. Front. 245–254 (2003). Biosci. (Sch. Ed.) 4, 831–839 (2012). 2. Li, G. & Reinberg, D. Chromatin higher-order structures and gene regulation. 32. Bonten, E. J. & d’Azzo, A. Lysosomal neuraminidase. Catalytic activation in Curr. Opin. Genet Dev. 21, 175–186 (2011). insect cells is controlled by the protective protein/cathepsin A. J. Biol. Chem. 3. Eberharter, A. & Becker, P. B. Histone acetylation: a switch between repressive 275, 37657–37663 (2000). and permissive chromatin. Second in review series on chromatin dynamics. 33. Smolewski, P. & Robak, T. The discovery and development of romidepsin for EMBO Rep. 3, 224–229 (2002). the treatment of T-cell lymphoma. Expert Opin. Drug Disco. 12, 859–873 4. Seto, E. & Yoshida, M. Erasers of histone acetylation: the histone deacetylase (2017). enzymes. Cold Spring Harb. Perspect. Biol. 6, a018713 (2014). 34. Minucci, S. & Pelicci, P. G. Histone deacetylase inhibitors and the promise of 5. Lin, H. Y., Chen, C. S., Lin, S. P., Weng, J. R. & Chen, C. S. Targeting histone epigenetic (and more) treatments for cancer. Nat. Rev. Cancer 6,38–51 (2006). deacetylase in cancer therapy. Med. Res. Rev. 26, 397–413 (2006). 35. Finnin, M. S. et al. Structures of a histone deacetylase homologue bound to the 6. Guan, J. S. et al. HDAC2 negatively regulates memory formation and synaptic TSA and SAHA inhibitors. Nature 401, 188–193 (1999). plasticity. Nature 459,55–60 (2009). 36. Nebbioso, A. et al. c-Myc Modulation and Acetylation Is a Key HDAC 7. Daneshpajooh, M. et al. HDAC7 is overexpressed in human diabetic islets and Inhibitor Target in Cancer. Clin. Cancer Res. 23, 2542–2555 (2017). impairs insulin secretion in rat islets and clonal beta cells. Diabetologia 60, 37. Lee, S. C., Bottaro, A. & Insel, R. A. Activation of terminal B cell 116–125 (2017). differentiation by inhibition of histone deacetylation. Mol. Immunol. 39, 8. Ishihama, K. et al. Expression of HDAC1 and CBP/p300 in human colorectal 923–932 (2003). carcinomas. J. Clin. Pathol. 60, 1205–1210 (2007). 38. Poole, C. J. & van Riggelen, J. MYC-master regulator of the cancer epigenome 9. West, A. C. & Johnstone, R. W. New and emerging HDAC inhibitors for and transcriptome. Genes 8, https://doi.org/10.3390/genes8050142 (2017). cancer treatment. J. Clin. Invest. 124,30–39 (2014). 39. Hwang, I. Y. et al. pVHL-mediated transcriptional repression of c-Myc by 10. Glaser, K. B. et al. Gene expression profiling of multiple histone deacetylase recruitment of histone deacetylases. Mol. Cells 33, 195–201 (2012). (HDAC) inhibitors: defining a common gene set produced by HDAC 40. Vo, B. T. et al. The interaction of Myc with Miz1 defines medulloblastoma inhibition in T24 and MDA carcinoma cell lines. Mol. Cancer Ther. 2, subgroup identity. Cancer Cell 29,5–16 (2016). 151–163 (2003). 41. Martina, J. A., Diab, H. I., Brady, O. A. & Puertollano, R. TFEB and TFE3 are 11. Pipalia, N. H. et al. Histone deacetylase inhibitor treatment dramatically novel components of the integrated stress response. EMBO J. 35, 479–495 reduces cholesterol accumulation in Niemann-Pick type C1 mutant human (2016). fibroblasts. Proc. Natl Acad. Sci. USA 108, 5620–5625 (2011). 42. Pastore, N. et al. TFEB and TFE3 cooperate in the regulation of the innate 12. Munkacsi, A. B. et al. An “exacerbate-reverse” strategy in yeast identifies immune response in activated macrophages. Autophagy 12, 1240–1258 (2016). histone deacetylase inhibition as a correction for cholesterol and sphingolipid 43. Sun, L. O. et al. Spatiotemporal control of CNS myelination by transport defects in human Niemann-Pick type C disease. J. Biol. Chem. 286, oligodendrocyte programmed cell death through the TFEB-PUMA axis. Cell 23842–23851 (2011). 175, 1811–1826 e1821 (2018). 13. Lu, J. et al. Histone deacetylase inhibitors prevent the degradation and restore 44. Kawauchi, D. et al. A mouse model of the most aggressive subgroup of human the activity of in Gaucher disease. Proc. Natl Acad. Sci. medulloblastoma. Cancer Cell 21, 168–180 (2012). USA 108, 21200–21205 (2011). 45. Klionsky, D. J. et al. A comprehensive glossary of autophagy-related molecules 14. Alam, M. S., Getz, M. & Haldar, K. Chronic administration of an HDAC and processes. Autophagy 6, 438–448 (2010). inhibitor treats both neurological and systemic Niemann-Pick type C disease 46. Mizushima, N., Yoshimori, T. & Levine, B. Methods in mammalian autophagy in a mouse model. Sci. Transl. Med. 8, 326ra323 (2016). research. Cell 140, 313–326 (2010). 15. Luzio, J. P., Hackmann, Y., Dieckmann, N. M. & Griffiths, G. M. The 47. Attisano, L., Silvestri, C., Izzi, L. & Labbe, E. The transcriptional role of Smads biogenesis of lysosomes and lysosome-related organelles. Cold Spring Harb. and FAST (FoxH1) in TGFbeta and activin signalling. Mol. Cell. Endocrinol. Perspect. Biol. 6, a016840 (2014). 180,3–11 (2001). 16. Mizushima, N. Autophagy: process and function. Genes Dev. 21, 2861–2873 48. Hassan, S., Sidransky, E. & Tayebi, N. The role of epigenetics in lysosomal (2007). storage disorders: uncharted territory. Mol. Genet. Metab. 122,10–18 (2017). 17. Singh, R. & Cuervo, A. M. Autophagy in the cellular energetic balance. Cell 49. d’Azzo, A., Machado, E. & Annunziata, I. Pathogenesis, emerging therapeutic Metab. 13, 495–504 (2011). targets and treatment in sialidosis. Expert Opin. Orphan Drugs 3, 491–504 18. Lim, C. Y. & Zoncu, R. The lysosome as a command-and-control center for (2015). cellular metabolism. J. Cell Biol. 214, 653–664 (2016). 50. Friedmann-Morvinski, D. & Verma, I. M. Dedifferentiation and 19. Slade, L. & Pulinilkunnil, T. The MiTF/TFE family of transcription factors: reprogramming: origins of cancer stem cells. EMBO Rep. 15, 244–253 (2014). master regulators of organelle signaling, metabolism, and stress adaptation. 51. Takahashi, K. et al. Induction of pluripotency in human somatic cells via a Mol. Cancer Res. 15, 1637–1643 (2017). transient state resembling primitive streak-like mesendoderm. Nat. Commun. 20. Sardiello, M. et al. A gene network regulating lysosomal biogenesis and 5, 3678 (2014). function. Science 325, 473–477 (2009). 52. Shin, H. J. et al. AMPK-SKP2-CARM1 signalling cascade in transcriptional 21. Haq, R. & Fisher, D. E. Biology and clinical relevance of the micropthalmia regulation of autophagy. Nature 534, 553–557 (2016). family of transcription factors in human cancer. J. Clin. Oncol. 29, 3474–3482 53. Fullgrabe, J. et al. The histone H4 lysine 16 acetyltransferase hMOF regulates (2011). the outcome of autophagy. Nature 500, 468–471 (2013). 22. Martina, J. A. et al. The nutrient-responsive transcription factor TFE3 54. Yi, C. et al. Function and molecular mechanism of acetylation in autophagy promotes autophagy, lysosomal biogenesis, and clearance of cellular debris. regulation. Science 336, 474–477 (2012). Sci. Signal 7, ra9 (2014). 55. Lee, J. Y. et al. HDAC6 controls autophagosome maturation essential for 23. Palmieri, M. et al. Characterization of the CLEAR network reveals an ubiquitin-selective quality-control autophagy. EMBO J. 29, 969–980 (2010). integrated control of cellular clearance pathways. Hum. Mol. Genet. 20, 56. Xie, H. J. et al. HDAC1 inactivation induces mitotic defect and caspase- 3852–3866 (2011). independent autophagic cell death in liver cancer. PloS ONE 7, e34265 (2012).

NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications 17 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11568-0

57. Jo, Y. K. et al. Up-regulation of UVRAG by HDAC1 inhibition attenuates gratitude to L. Attisano for her gift of the flag-FOXH1 plasmid, R. Puertollano for 5FU-induced cell death in HCT116 colorectal cancer cells. Anticancer Res. 38, providing the Tfeb/TFe3 dKO MEFs, K. Freeman for the neuroblastoma cells, M. Kundu 271–277 (2018). for the LC3-GFP plasmid. A.d’A. holds the Jewelers for Children Endowed Chair in 58. Chauhan, S. et al. ZKSCAN3 is a master transcriptional repressor of Genetics and Gene Therapy. This work was supported, in part, by NIH grants autophagy. Mol. Cell 50,16–28 (2013). GM104981, and CA021764 (A.d’A.), P01CA096832, and P30 CA02165 (M.F.R.), the 59. Seok, S. et al. Transcriptional regulation of autophagy by an FXR-CREB axis. Assisi Foundation of Memphis (A.d’A.), the American Lebanese Syrian Associated Nature 516, 108–111 (2014). Charities (ALSAC) (A.d’A.), and by a grant from the Deutsche Forschungsgemeinschaft 60. Zhao, J. et al. FoxO3 coordinately activates protein degradation by the (Emmy Noether Research Group, WO 2108/1-1) (E.W.). autophagic/lysosomal and proteasomal pathways in atrophying muscle cells. Cell Metab. 6, 472–483 (2007). Author contributions 61. Mammucari, C. et al. FoxO3 controls autophagy in skeletal muscle in vivo. – I.A. designed, performed, and analyzed most of the experiments. D.v.d.V. performed all Cell Metab. 6, 458 471 (2007). the RT-qPCR, western blot analyses and assembled the figures. E.W. performed com- 62. Zhao, Y. et al. Cytosolic FoxO1 is essential for the induction of autophagy and – putational analyses of ChIP datasets. D.F. performed computational analyses of ChIP tumour suppressor activity. Nat. Cell Biol. 12, 665 675 (2010). datasets, gene expression and correlation analyses of publicly available datasets. G.N. 63. Talchai, C., Xuan, S., Kitamura, T., DePinho, R. A. & Accili, D. Generation of performed, analyzed, and deposited the microarray data. H.T. scored and analyzed tissue functional insulin-producing cells in the gut by Foxo1 ablation. Nat. Genet. 44, – microarrays and patient-derived xenografts. E.M. performed immunohistochemistry, 406 412, S401 (2012). western blot analyses and data acquisition. R.M. performed enzymatic activity assays and 64. Hoodless, P. A. et al. FoxH1 (Fast) functions to specify the anterior primitive fi fl – cultured primary skin broblasts. Y.C. performed immuno uorescence analyses and co- streak in the mouse. Genes Dev. 15, 1257 1271 (2001). immunoprecipitation analyses. M.F.R. provided the mouse group 3 medulloblastoma 65. Saunders, A. et al. The SIN3A/HDAC corepressor complex functionally – cells and intellectual input. J.A.W. and L.E.F. performed western blots analyses. X.Q. cooperates with NANOG to promote pluripotency. Cell Rep. 18, 1713 1726 purified in house antibodies. M-J.H. reprogrammed the dermal fibroblasts into hIPSCs (2017). and performed proteomic analyses. G.C.G. provided reagents, interpreted data, and 66. Suva, M. L., Riggi, N. & Bernstein, B. E. Epigenetic reprogramming in cancer. ’ – edited the paper. A.d A. designed experiments, analyzed data, and supervised the study. Science 339, 1567 1570 (2013). I.A. and A.d’A. conceived the project and wrote the manuscript. 67. Soufi, A., Donahue, G. & Zaret, K. S. Facilitators and impediments of the pluripotency reprogramming factors’ initial engagement with the genome. Cell 151, 994–1004 (2012). Additional information 68. Takahashi, K. & Yamanaka, S. Induction of pluripotent stem cells from mouse Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- embryonic and adult fibroblast cultures by defined factors. Cell 126, 663–676 019-11568-0. (2006). 69. Takahashi, K. et al. Induction of pluripotent stem cells from adult human Competing interests: The authors declare no competing interests. fibroblasts by defined factors. Cell 131, 861–872 (2007). 70. Gabay, M., Li, Y. & Felsher, D. W. MYC activation is a hallmark of cancer Reprints and permission information is available online at http://npg.nature.com/ initiation and maintenance. Cold Spring Harb. Perspect. Med. 4, https://doi. reprintsandpermissions/ org/10.1101/cshperspect.a014241 (2014). 71. Felsher, D. W. MYC inactivation elicits oncogene addiction through both Peer review information: Nature Communications thanks Joe Delaney and other tumor cell-intrinsic and host-dependent mechanisms. Genes Cancer 1, anonymous reviewer(s) for their contribution to the peer review of this work. Peer 597–604 (2010). reviewer reports are available. 72. Ecker, J. et al. Targeting class I histone deacetylase 2 in MYC amplified group 3 medulloblastoma. Acta Neuropathol. Commun. 3, 22 (2015). Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in 73. Pei, Y. et al. HDAC and PI3K antagonists cooperate to inhibit growth of published maps and institutional affiliations. MYC-driven medulloblastoma. Cancer Cell 29, 311–323 (2016). 74. Izzi, L. et al. Foxh1 recruits Gsc to negatively regulate Mixl1 expression during – early mouse development. EMBO J. 26, 3132 3143 (2007). Open Access This article is licensed under a Creative Commons 75. Truax, A. D. & Greer, S. F. ChIP and Re-ChIP assays: investigating Attribution 4.0 International License, which permits use, sharing, interactions between regulatory proteins, histone modifications, and the DNA – adaptation, distribution and reproduction in any medium or format, as long as you give sequences to which they bind. Methods Mol. Biol. 809, 175 188 (2012). appropriate credit to the original author(s) and the source, provide a link to the Creative 76. te Vruchte, D. et al. Relative acidic compartment volume as a lysosomal Commons license, and indicate if changes were made. The images or other third party storage disorder-associated biomarker. J. Clin. Invest. 124, 1320–1328 material in this article are included in the article’s Creative Commons license, unless (2014). indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from Acknowledgements the copyright holder. To view a copy of this license, visit http://creativecommons.org/ We are indebted to D. Peach for his stimulating inputs. We thank A. Ballabio for critical licenses/by/4.0/. reading of the manuscript; S. Frase, L. Horner for the electron microscopy images and quantification; R. Ashmun for the FACS analyses; C. Qu for help with motif analyses; and F. Harwood for assistance with the transfection protocols and reagents. We extend our © The Author(s) 2019

18 NATURE COMMUNICATIONS | (2019) 10:3623 | https://doi.org/10.1038/s41467-019-11568-0 | www.nature.com/naturecommunications